Re: IXF and SDMXHD
Bob Jolliffe
Sorry for delayed response.
On 19 June 2011 04:27, Alvin Marcelo <alvin....@gmail.com> wrote:
> Hi Bob, Knut,
>
> We just came out of the 2nd Connectathon. (Documentation of the first is
> attached). In the 2nd Connectathon, I proposed that we look at the WHO IMR
> and see if we can:
>
> 1. use it to convert paper indicator definitions into electronic
> 2. use it to structure local indicator definitions into an ISO standard
> 3. share it inside the IMR for others to see and emulate
> 4. print into PDF for dissemination and manualization
> 5. export as raw SDMX-HD files for further customization by developers with
> local codelists (Connectathon3)
> 6. demonstrate (very briefly) how field reporting systems can comply with
> SDMX-HD to submit compliant files
>
> We're still writing up the documentation for C2 but I'd like to share some
> personal insights and look forward to your thoughts as well:
>
> 1. SDMX-HD is heavy! You need to send the whole zip folder back and forth
> the whole time?
> 2. What is IXF and when is it better than SDMX-HD?
> 3. Is there an IMR (or similar portal) for IXF?
>
> My take from the C2 was:
>
> 1. we can use IMR to define country level indicators using ISO SDMX-HD
> format (by itself, this already brings us to higher level by adopting good
> practice)
> 2. we can use external URN (namespaces) for codelists (good practice)
> 3. extend the SDMX-HD with our local codes (to make it relevant
> sub-nationally)
> 4. BUT allow vertical report submission to use IXF (eg, community clinic and
> provincial health office) and use SDMX-HD for horizontal exchange (eg, bet
> Ministry of Health and Ministry of Education)
>
> What do you think?
Agree wholeheartedly with 1 to 3. But I *really* don't think we want
to go resurrecting IXF at this stage. In fact I am pretty sure we
have retired support for it in DHIS2.
What you want to do is to transport tabular data from one system to
another within the country. SDMX-HD is perfectly well suited for that
and you really don't want to throw in redundant standards to the mix..
But sending the full zip package backwards and forwards is indeed
heavy and I would not recommend doing this all the time. You really
want to do this when you are sharing DSD and rich metadata between
systems. But for routine reporting, after the DSD consensus has been
established, it is sufficient to transmit SDMX data messages - we have
found that for routine, periodic reporting the cross-sectional dataset
is most appropriate for that. And there may even be instances where a
CSV message in which the definition of columns is determined by a DSD
can also be appropriate. Particularly when the amount of data is
large and/or the pipes you need to send it through are narrow. SDMX
provides for this and I can foresee cases, particularly in bandwidth
constrained environment like we see often in Africa, that we might
want to write this into SDMX-HD. So yes the full SDMX-HD zip package
is heavy. And it needs to be to describe the full reference
information for indicators and all their disaggregations. But the zip
packaging is not a mandatory feature for all use cases. There is
nothing in the standard to prevent the exchange of simple SDMX-HD data
messages.
The important thing for in country reporting is that you probably want
to transport dataelement data values rather than indicator values.
This is because you usually will want to aggregate and calculate the
indicator values at different levels (district, province, national
etc). For this reason, you need the "raw" values before information
is lost by dividing by denominators. Having mortality rates as
percentage values from twenty districts does not allow you roll those
up to province level, for example. For that you need the counts.
SDMX-HD caters for this kind of message. Specifically the OBS_VALUE
element is required to have either an INDICATOR attribute or a
DATAELEMENT attribute. For in-country data reporting you will most
likely use the former rather than the latter. In fact I suspect the
use cases for data messages containing indicator values is are
actually far fewer - mostly concerned with publication, where for
example, an SDMX-HD dataset might be transformed and displayed on a
web page, or for analysis where the indicator values might be pivoted
against the dimensions to explore relationships in the data.
Given that you are on the road to doing 1 to 3 above (which is really
excellent in that it establishes a focus for creating common codes
which derive their legitimacy at a country/international level) it
would be a terrible waste if you did not choose to reuse that work by
establishing SDMX-HD as your standard reporting form of dataelement
values within the country. The codelists which you create for your
national level indicator metadata should be the same ones you use
throughout. The use of SDMX for this is not in competition to its
grander use at the level of defining national indicators. This is
precisely the way we gradually harmonize what Patrick, Xavier and co
have been doing with the IMR, and what we have been doing between
systems at the grassroots.
>
> Also, do you think I should be having separate exchanges bet you and
> Patrick/Xavier? Or can we just have one big thread for all of us?
Absolutely. To which end I have copied this to the sdmx-hd mailing
list which is still hosted in the google group. Patrick, Xavier,
Knut and many others are on this list. I also earlier today tried to
send you an invite to the group. Did you get it? Unfortunately I am
not sure exactly who set this up or who has the rights to accept new
members. I am sure Patrick will know better.
As a general observation, I think we do need to put back some effort
into the governance of SDMX-HD. I have been meaning to address myself
to this at greater length, but not got to it yet. Despite all the
warts and imperfections, the value that SDMX-HD brings potentially,
which I think IXF does not, is that it it can and should be an open
standard which is open and transparent and driven by an open community
of implementors and users - I'll elaborate on this later, but a
significant aspect should be to make use of an open and archived list
(of which google groups is the best we have for now).
It seems there has been lots of interesting and useful things going on
in Phillipines. Similarly there has been some useful work done
between iHRIS and DHIS in Zanzibar and ongoing work with the OpenMRS
derived hospital system in India. So in response to your request re
single thread, I would certainly encourage that we try to have these
discussions on list.
Sorry all I have time for now :-) Must get back to work ...
Regards
Bob
>
>
> alvin
>
>
>
>
>
>
> --
> Alvin B. Marcelo, MD, FPCS
> www.alvinmarcelo.com
> Voicemail: +1-301-534-0795)
> GPG 0x99CBC54C
>
>
>
Bob Jolliffe
Correcting myself on re-reading my mail. 'tis the DATAELEMENTS you
want for routine reporting ...
Timothy Cook
Thanks for copying this to the list.
Cheers,
Tim
> --
> You received this message because you are subscribed to the Google Groups "SDMX-HD (Health Domain)" group.
> To post to this group, send email to sdm...@googlegroups.com.
> To unsubscribe from this group, send email to sdmx_hd+u...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/sdmx_hd?hl=en.
>
>
--
================
Timothy Cook, MSc
Visit my blog at: http://hiiacw.blogspot.com/
LinkedIn Profile:http://www.linkedin.com/in/timothywaynecook
Skype ID == timothy.cook
Academic.Edu Profile: http://uff.academia.edu/TimothyCook
Whitaker, John Patrick
BTW, the IXF is a UNAIDS exchange format for the CRIS application, a precursor to the SDMX-HD.
Patrick
Patrick Whitaker
eHealth and Informatics Unit
Health Statistics and Informatics Department World Health Organization Geneva, Switzerland
Tel. direct: +41 22 791 1372
Mobile: +41 79 633 7763
E-mail: whit...@who.int
Health Informatics Wiki
http://www.hiwiki.org
SDMX-HD Standard
http://www.sdmx-hd.org
WHO Indicator and Measurement Registry
http://apps.who.int/gho/indicatorregistry
--
Bob Jolliffe
> Just to clarify, 'raw' can still mean summary data. The IMR supports percentage and ratio indicator definitions which have numerator and denominator defined as indicators themselves. CDW at national level may still be out of reach in many places.
Agree. I wasn't using raw in the sense of clinical data. More in the
sense of raw counts - the kind of dataelement which would typically be
found on (paper) monthly reporting forms used by facilities. The
number of malaria cases, the number of women enrolled in ANC, number
of vaccinations given etc.
To produce electronic SDMX-HD versions of these reports, a DSD needs
to be created which encapsulate these routinely reported dataelements.
Then datasets can be brought into national/provincial/district data
warehouses which conform to this. Assuming the same facility and
disaggregation codelists have been used top-to-bottom these can then
be transformed into indicator reports for publication/analysis.
Something which is not quite clear to me. Who or what will be
consuming these indicator reports in Phillipines?
Regards
Bob
Alvin Marcelo
Thanks for the long reply Bob...appreciate it...
Well in the Philippines, we have the "usual" indicators. I don't really see what other countries have so "usual" to us are:
maternal mortality ratio
child mortality under 5 years
TB prevalence rate
case detection rate
etc
My concern was that I was seeing a lot of ratios and percentage reports for org units as small as 5000 population! Is that correct at all?
My take was that all org units, at any level below national, should be reporting counts (percentages)...Or if forced to submit percentages -- to include the raw numerator and denominator (as prescribed in the SDMX-HD DSD)...
eloy
r.fri...@mindspring.com
Are the Philippines using the OpenMRS/DHIS2 combination? If not, what? If so, you may be interested in a GSOC project I am mentoring this summer to integrate DHIS2 via the reporting framework.
Saludos, Roger
To unsubscribe from this group and stop receiving emails from it, send an email to sdmx_hd+u...@googlegroups.com.
To post to this group, send email to sdm...@googlegroups.com.
Visit this group at http://groups.google.com/group/sdmx_hd.
For more options, visit https://groups.google.com/groups/opt_out.
Alvin B. Marcelo
We're back to selecting the schema. The openmrs-dhis2 was an experiment which worked (openmrs can submit data to dhis2). But I wanted to offer a wider set of options and have a multistakeholder panel select.
Of course there will now be a bias to what already works but I'd like to see also how we can leverage the Indicator and Measurement Registry.
Steven Uggowitzer
Carl Fourie
Sent from my iPad
Alvin Marcelo
Dominic Atweam K
Sent from my iPhone
r.fri...@mindspring.com
Folks --
I am attaching some e-mails from Phillipe Boucher about work going on at WHO. They seem to be moving to XMart as their tool of choice. GHO refers to Global Health Observatory. From what I have been able to glean, the WHO Regions and programs each have DBs of data for countries in their region and Geneva is looking to consolidate them in the GHO. Therefore, their interest in the clinical-aggregate connection is low. The IMR appears to be withering from lack of support. Some people here believe WHO is not the right host for IMR. There was an attempt to discuss these issues in April or May in the OpenMRS space, but a critical mass of people never emerged.
For the GSOC project, we decided to move away from SDMX-HD because our primary goal was to simplify OpenMRS to DHIS2 data exchange. We decided to focus on putting the indicator calculation into the reporting framework (as opposed to writing SQL queries against the OpenMRS DB as is done by HISP) and on solving the differences in representing disaggregation (with OpenMRS like SDMX-HD using dimensions while DHIS2 uses a single dimension representing multiple category combinations).
Saludos, Roger
Bob Jolliffe
--
BOUCHER, Phillipe Jean-Pierre
Hi,
Just to clarify, it’s not the Indicator Metadata Registry (IMR) that is withering at WHO (in fact we’re trying to expand it’s use J ) it’s the support for SDMX-HD – we don't have technical resources available to keep it moving forward or lead it. If the community were to take it over, we could probably participate, most likely in the context of its applicability to the Global Health Observatory, but we probably couldn't do much else
Philippe
From: sdm...@googlegroups.com [mailto:sdm...@googlegroups.com]
On Behalf Of Bob Jolliffe
Sent: 06 August 2013 15:42
To: sdm...@googlegroups.com
Subject: Re: IXF and SDMXHD
Hi Roger