MCP App Store
by KeenEthics
RxIQ icon

RxIQ MCP Server for ChatGPT

By Andrew Charlwood

Overview

RxIQ answers questions about NHS primary care prescribing, from national level down to individual GP practices. It is built for medicines optimisation teams, pharmacists, commissioners, analysts and anyone researching how medicines are prescribed and paid for in England. Ask about any medicine, BNF section or class of drugs, and about any integrated care board (ICB), sub-ICB area, primary care network (PCN) or GP practice. RxIQ can: - Query prescribing by medicine, place and month: items, quantity and cost, totals and per 1,000 patients, ranked and split by area or product, for GP or non-GP prescribers. - Summarise a medicine: national spend and trend, the ICBs that prescribe most per patient, and its current Drug Tariff price. - Profile a place: list size, spend and year-on-year growth for an ICB, PCN or practice. - Report clinical measures (antibiotics, opioids, respiratory, diabetes, safety and more), with each place's value, England's value and its rank. - Show Drug Tariff price history and price concessions for a pack, with the estimated extra cost to the NHS. - Give an ICB's finance position against national benchmarks, its forecast and savings targets. - Return the headline and summary of an ICB's monthly prescribing intelligence report, with a link to the full report. - Draft a dashboard on rxiq.co.uk from a description, with a link to view it and save a copy. What RxIQ adds to the raw NHSBSA open data: a fast query engine over five years of monthly practice-level prescribing, clinical measures ranked across all 36 ICBs, Drug Tariff and concession history joined to prescribing volumes, need-adjusted finance benchmarks and forecasts, and monthly ICB intelligence reports. Every answer links to the matching page on rxiq.co.uk so figures can be checked. RxIQ holds aggregate prescribing data published by the NHS Business Services Authority. It holds no patient-level data and does not give clinical advice. Spend figures count GP prescribing unless non-GP prescribing is asked for. No account is needed.

How to connect

  1. ChatGPT: Open the RxIQ listing in the ChatGPT app directory and choose Connect. ChatGPT listing RxIQ exposes 16 tools here.

Tools

data_status

ChatGPT
What RxIQ's data covers: the latest month of NHSBSA prescribing data, the first and last months that can be queried, the month registered patient counts start, the date of the organisation map, the caveats that apply to every figure (GP-only spend, concessions inside NIC, the two-month lag, how measures are defined), and in caveat_details the full text of the caveats other tools attach in short (per 1,000 patient figures, ADQ, below-ICB measures, non-GP estimates). Fits checking how current the data is before answering.

draft_dashboard

ChatGPT
Makes a draft RxIQ dashboard from a whole spec and returns its view link and an edit key. The draft is checked as the dashboard builder checks a saved dashboard, against the widget catalogue. Opening the view link shows the dashboard, where anyone holding the link can view it and, signed in, save a copy to their RxIQ account; the edit key changes the draft. Drafts expire 7 days after their last edit. Fits making a dashboard for a project, such as a switch or a measure to improve, without signing in.

edit_draft

ChatGPT
Changes a dashboard draft by screen and widget id: adds widgets and screens, renames screens, and changes widgets' settings, titles, sizes and places and the dashboard's name, theme, area, period and medicines. It never removes anything. Returns only what changed (the dashboard's fields, and each screen and widget added or changed, in full), the widget and screen counts, the view link, the expiry and any notes; get_draft gives the whole draft. A copy someone already saved from the draft stays as it was.

get_draft

ChatGPT
Reads a dashboard draft back as it now is, in the shape the draft tools answer with: its name, line, screens and widgets with their ids, settings and places, its view link and its expiry. Takes the edit key, or the view link alone; it changes nothing, and gives the edit key back only when it was the edit key it was given. Fits checking a draft before editing it, or picking up a draft made earlier in the conversation.

get_finance_position

ChatGPT
An ICB's (or England's) GP prescribing finance position from RxIQ's finance tool: spend over the latest 12 months, cost per registered patient (the intelligence report's Cost per patient), cost per need-weighted patient against three benchmarks (national average, median ICB, lowest-cost quarter) with the gap to each, the spend forecast by financial year against each year's budget baseline with its range, and savings targets that close half the gap to the national average over two years from next financial year (each year's saving is cumulative: it includes the earlier years'). Each benchmark says which ICBs it stands for. Fits questions about budgets, overspend, forecasts and savings targets.

get_intelligence_report

ChatGPT
The summary of the published RxIQ prescribing intelligence report for an ICB (or England): its headline, standfirst, section titles and how many decisions it puts to the ICB, with the month and a link. The full report (each section's analysis, tables, actions and practice detail) is read at the link, not here. Fits questions about what an ICB's latest report covers and where to read it.

get_measures

ChatGPT
Figures for one measure, or every measure in one theme, at one place: England, an ICB, a sub-ICB location, a PCN or a GP practice. Gives the latest month's value and the latest 12 months against the 12 before with the change, England's figures beside it, and a link to the measure's page. Ranks use two windows, which rank_windows names once, place for the place's own ranks and icb for its ICB's: an ICB is ranked of the 36 ICBs on the latest month's value, as the measure page ranks them (most favourable first, '=' marking a joint rank); a sub-ICB location, PCN or practice carries its ICB's figures too and is ranked among its ICB's locations, PCNs or practices on its pooled 12 months (a practice under 1,000 patients or without a full year is shown but not ranked). Each rank's outer_group says where it falls: most_favourable_4_of_36_icbs or least_favourable_4_of_36_icbs for an ICB, most_favourable_tenth_in_icb or least_favourable_tenth_in_icb below one, else middle; null on a measure with no better direction, and below an ICB when fewer than 10 places are ranked, where a tenth means nothing and the rank alone says it. An unknown id or theme is answered with the closest. Fits questions about a place's antibiotic, opioid, safety or other prescribing quality measures.

get_medicine

ChatGPT
One medicine's national picture from RxIQ's medicine page: its name and BNF placement, GP prescribing items and spend in England by month for the latest 12 months (the period; period.months lists them, oldest first, and england.monthly holds arrays, items and spend, whose nth value is period.months' nth month) with the period's totals against the 12 months before, the 10 ICBs with the most items per 1,000 registered patients (items and spend per 1,000 over the period, _in_period, and for the mean month, _per_month), and each pack's price in the latest Drug Tariff with its month and, where the latest complete concession month had a concession, the concession price beside the tariff price in force that month. A medicine whose list price misstates NHS cost (tirzepatide) carries price_notes with sources. Fits questions about one medicine's use, cost, trend or price; query_prescribing fits comparisons and other splits.

get_place_profile

ChatGPT
Descriptive facts about England, an ICB, a sub-ICB location, a PCN or a GP practice: its name and parents (practice, PCN, sub-ICB, ICB), registered patients in the latest month, and its prescribing (items, actual cost and NIC) by month for the latest 12 months (the period; period.months lists them, oldest first, and monthly holds arrays, items, actual_cost and registered_patients, whose nth value is period.months' nth month), with the period's totals (_in_period), items and cost per 1,000 patients over the period (_in_period) and for the mean month (_per_month), and growth as the intelligence report measures it: actual cost in the latest 3 months against the same 3 months a year earlier. Fits questions about a place's size and overall prescribing; query_prescribing splits it by medicine.

get_tariff_and_concessions

ChatGPT
Drug Tariff packs with their price at each change since 2021 (tariff_history, rows of [month, tariff_price]; tariff_price is the latest, unchanged since tariff_price_since), every month a price concession applied with the concession price beside the tariff price in force that month, and RxIQ's estimate of England's cost above tariff that month (concession_months, rows of [month, concession_price, tariff_price_that_month, england_cost_above_tariff]; a month past the latest prescribing data is estimated, and its row has a fifth value, the months of volume it was priced on). columns names each row's values once. Finds packs by name, pack code or product code; with a name, lists up to 10 matching packs in matches and details the first three. A pack whose list price misstates NHS cost (tirzepatide) carries price_notes with sources. Fits questions about a generic's price, price rises, and concession history.

list_dashboard_widgets

ChatGPT
The widgets an RxIQ dashboard can hold, as the dashboard builder offers them. Asked with nothing, an index: the dashboard's own fields (area, medicines, period, theme, screens and the grid), each category with its widgets' names, and the starter dashboards with what each is for. Asked for a category, only that category's widgets in full: what each shows, its default and least sizes, and each setting's allowed values and default, a setting several widgets share described once in common_settings. Asked for a starter with an area, only that starter's layout, as a ready dashboard spec. These are widget types, not data. Fits planning a dashboard to make with draft_dashboard.

list_measures

ChatGPT
RxIQ's clinical measures (most follow OpenPrescribing's definitions) and the themes they sit in: each measure's id, name, theme, unit, and which direction is better (with a note where that reads oddly, as on a generic share of a medicine prescribed by brand). With neither a theme nor a medicine, the measures are listed under their themes, and lower is better on every one that names no direction of its own. When a theme or medicine narrows the list, each measure also carries its notes and clinical context where it has them, and its national target where it has one (few do: most measures have none, and target is then left out). Fits finding which measures exist for a topic or a medicine before get_measures reads their figures.

query_prescribing

ChatGPT
Counts NHS primary care prescribing for chosen medicines and places over chosen months, split into rows by medicine level (BNF chapter to brand) or place level (England to practice), ranked by items, actual cost, net ingredient cost, quantity or ADQ. With one level, returns up to limit rows (default 20). With two, limit (default 10) caps the first-level rows and child_limit (default 5) each one's second-level rows: every first-level row in the limit holds its own top children in children, with children_total and children_truncated, so a large first row never crowds out the next. To keep answers small, limit × (child_limit + 1) is at most 60, and with include_monthly those rows × months at most 720. Each row has its totals for the period (items, actual cost, NIC, quantity by unit, ADQ where defined, and for GP rows items and cost per 1,000 registered patients, both over the period (_in_period) and for the mean month (_per_month, left out for a one-month period, where it is the same), rates below 1 to two significant figures, and ADQs per 1,000 patients per day), with the selection's total by month, the months covered (period.months lists them, oldest first; a monthly holds arrays, items and actual_cost, whose nth value is period.months' nth month; months outside the data are held to its range, and the summary says so; list sizes begin in November 2024, so per-1,000 figures cover only the months since, which period.per_1000 names, with their month_count, when they are fewer), the medicines included and excluded by name, whether first-level rows were cut (truncated, of total_rows), and a link to the same view in RxIQ's Analyse. Rows, at either level, at a place level coarser than an area asked hold only the areas asked: such a row's name says so ("Kings Lynn PCN (Vida Healthcare only)", "(2 of its 9 practices)"), covers counts its open practices included, and with one level the areas asked are its children, adding up to it. An area inside another asked is marked inside and counted once. Ranked by quantity, rows rank by the largest unit group (ranked_by_unit_group), each keeping its quantity in every unit group. ADQ (NHSBSA's average daily quantity) is defined for some medicines only: ranking by ADQ where none of the medicines has one is refused, and where only some have one, adq_coverage gives the share of items with an ADQ and names the chemicals without, which the ADQ figures leave out; ranked by ADQ, rows without one come after the ranked rows, most actual cost first, marked no_adq. Non-GP items are estimated shares, so can be fractions, and non-GP prescribers have no sub-ICB location or PCN. Fits questions such as which ICBs prescribe most of a medicine, how a practice's prescribing splits by product, or a trend over time. Medicine and place codes come from search.

remove_from_draft

ChatGPT
Removes any number of widgets and screens (a screen with its widgets) from a dashboard draft at once, by id. The draft keeps at least one screen. Returns the ids removed, the widget and screen counts left, the view link, the expiry and any notes; get_draft gives the whole draft.

replace_draft

ChatGPT
Replaces a dashboard draft's whole layout with a new spec, keeping its view link and edit key. Returns what differs from the draft it replaced (the dashboard's fields, and each screen and widget added or changed, in full, and the ids removed), the widget and screen counts, the view link, the expiry and any notes; get_draft gives the whole draft.

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/rxiq"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/rxiq)