Generate bespoke plots for maternity, gynae and paeds

[matt-dray]

Output reports typically use the default codebase we've built up, regardless of scheme. What if we needed to focus on maternity, gynae and paeds? What changes would be needed to the code? What results data and filters should be used? What isn't possible?

Definitions for this purpose:

• 'maternity', which is already used in the point of delivery (POD) filters in the outputs app
• 'gynae', i.e. treatment specialties (tretspef) 502 and 503 (noting that 503 is grouped into one of the 'other' categories in tretspef_grouped)
• 'paeds', i.e. ages 0 to 17

Possibilities for each chart type in the reports:

• Waterfalls (fig 9.4) and change-factor (8.2, 8.3, 8.4, 8.5, 8.6) charts: use existing POD filter for maternity; gynae and paeds not possible becuase tretspef and age aren't available in step counts.
• Beeswarms (9.1, 9.6, 9.8, 9.10) and S-curves (9.2, 9.7, 9.9, 9.11): use existing POD filter for maternity; replace r$results$default data in get_model_run_distribution() with filtered r$results$tretspef for gynae; with filtered r$results$age for paeds.
• Summary table (9.3): filter to retain the maternity POD only; replace r$results$default data in get_principal_high_level() with filtered r$results$tretspef for gynae; with filtered r$results$age for paeds; alternatively, consider using the results in detail table, which contains a split for tretspef (though its actually tretspef_grouped, which doesn't expose 503) and age.
• LoS table (9.5): filter to retain the maternity POD only; filter existing r$results$tretspef+los_group in mod_principal_summary_los_data() for gynae; not possible for paeds becuase there's no age+los_group in the results data.

[yiwen-h]

For LoS table we could do a custom age+los_group using detailed results but would be extra work.

Actually just running detailed_results in nhp_products might produce what you need? without extra work

[matt-dray]

Thanks YiWen. You may be right about using the detailed results for a paeds LoS table, becuase it has los_group, age_group and mean (principal) columns. Where would we get the corresponding baseline values from, which are needed to calculate change? Detailed baseline doesn't split by age iirc.

Some further notes (mostly for my benefit):

1. Dataset selection isn't controlled by an argument, e.g. r$results$default is hardcoded early in a sequence of functions that produces beeswarms, so we'd have to add an argument like data_type = "tretspef" that gets passed through these functions (or maybe control data ingestion via config or options()?).
2. {reskit} has functions to produce some of these charts/tables, but I think we should use what we have in the repo for now, given that this repo handles all model versions (not just latest).
3. We'll script this task separately from the main codebase for now and then think about integrating it later, noting that it will add complexity to the main codebase and will deviate it further from nhp_outputs and {reskit}.

[yiwen-h]

We could get baseline values from the raw data but it would be a custom script, so a pain to do. Again it goes back to "how important is this" and "how much time do we want to spend on it"

[matt-dray]

Re point (1) above: I've gone with options() approach, so for e.g. beeswarms/s-curves we can edit get_model_run_distribution() (the first in a series of functions required to produce the plots) to add a switch():

results_data <- switch(
  getOption("filter_type", default = NA_character_),
  "gynae" = r$results$tretspef |> dplyr::filter(tretspef %in% c("502", "503")),
  "paeds" = r$results$age |> dplyr::filter(dplyr::between(age, 0, 17)),
  r$results$default
)

This avoids having to e.g. cascade a new argument through all functions in the plot-producing pipeline.

So, for now, the process would be:

1. Load all existing functions.
2. Overwrite get_model_run_distribution() with a version that contains the switch.
3. Set e.g. options(filter_type = "gynae") and run the modified function.

Like:

options(filter_type = "gynae")
plots_activity_distribution_gynae <-
  prepare_all_activity_distribution_plots(r_primary, site_codes)
options(filter_type = NULL)  # reset

[matt-dray]

@jacgrout: I think I've got everything I mentioned was possible in the description above. It's in a script in the draft PR #55. I'm going to book in a chat for us to talk it through.

[jacgrout]

[like] GROUT, Jacqueline (NHS MIDLAND... reacted to your message:

…

________________________________ From: Matt Dray ***@***.***> Sent: Friday, January 9, 2026 5:32:33 PM To: The-Strategy-Unit/nhp_output_reports ***@***.***> Cc: GROUT, Jacqueline (NHS MIDLANDS AND LANCASHIRE COMMISSIONING SUPPORT UNIT) ***@***.***>; Mention ***@***.***> Subject: Re: [The-Strategy-Unit/nhp_output_reports] Examine the possibility of reporting for maternity, gynae and paeds (Issue #54) This message originated from outside of NHS.net Connect. Please do not click links or open attachments unless you recognise the sender and know the content is safe. [https://avatars.githubusercontent.com/u/18232097?s=20&v=4]matt-dray left a comment (The-Strategy-Unit/nhp_output_reports#54)<#54 (comment)> @jacgrout<https://github.com/jacgrout>: I think I've got everything I mentioned was possible in the description above. It's in a script in the draft PR #55<#55>. I'm going to book in a chat for us to talk it through. — Reply to this email directly, view it on GitHub<#54 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AYVITYN335KND4AXUSFBCTD4F7Q3DAVCNFSM6AAAAACQ3NCWDKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTOMRZHEYTCMZWGE>. You are receiving this because you were mentioned.Message ID: ***@***.***> ************************************************************************************** ****************************** This message may contain confidential information. If you are not the intended recipient please: i) inform the sender that you have received the message in error before deleting it; and ii) do not disclose, copy or distribute information in this e-mail or take any action in relation to its content (to do so is strictly prohibited and may be unlawful). Thank you for your co-operation. NHS.net Connect is the secure email, collaboration and directory service available for all NHS staff in England. NHS.net Connect is approved for exchanging patient data and other sensitive information with NHS.net Connect and other accredited email services. For more information and to find out how you can switch visit Joining NHS.net Connect – NHS.net Connect Support<https://support.nhs.net/article-categories/joining-nhsmail/>

[matt-dray]

@jacgrout: you asked about getting p10s and p90s for the S-curves. Here's a simple example. In the dev/mat-gynae-paeds.R script of #55, run the 'set up' and 'helper functions' sections and also the bespoke get_model_run_distribution() function from the 'generate beeswarms and S-curves' section. The following snippet then gives the p10s and p90s for IP bed days, for example:

atmpo <- read_atmpo() # table of activity type/pod/measure permutations
at <- "inpatients"
m <- "beddays"
pods <- atmpo |>
  dplyr::filter(activity_type == at, measure == m) |> 
  dplyr::pull(pod)

options(results_type = NULL) # ensure this is unset for the overall values
get_ecdf_quantiles(r_primary, site_codes, at, pods, m) # no gynae/paeds filtering
options(results_type = "gynae")
get_ecdf_quantiles(r_primary, site_codes, at, pods, m)
options(results_type = "paeds")
get_ecdf_quantiles(r_primary, site_codes, at, pods, m)
options(results_type = NULL) # unset

For maternity you could set the argument pods = "ip_maternity_admission" directly in get_ecdf_quantiles() and unset the results_type option with options(results_type = NULL).