167x Filetype PDF File size 0.09 MB Source: www.cso.ie
Implementation of IRIS software for the automated coding of deaths in Ireland 1 Implementation of IRIS software for the automated coding of deaths in Ireland Background coding as MMDS is no longer being supported by the US Centre for Disease Control who have moved to join The production of cause of mortality statistics requires the IRIS development core group. the selection of an underlying cause of death from death IRIS software contains language-dependent tables that certificates. This underlying cause of death is defined as can be developed to suit individual jurisdictions e.g. (a) the disease or injury which initiated the train of morbid we can include common phrases used in Irish death events leading directly to death, or (b) the circumstances certificates which may not be included in the standard of the accident or violence which produced the fatal English dictionary released with the IRIS package. injury. As part of this process the CSO used to use the US Centre of Disease Control-developed Medical Mortality The IRIS user interface is more fluid, intuitive and user- Data System (MMDS) suite of software but switched to friendly than the older MMDS system. using IRIS from 2018 onwards. While not altering the classifications per se, a change The IRIS MUSE component provides detailed in coding software can affect the assignment of the explanations on how the system arrived at the multiple underlying cause of death codes. Therefore, as part and underlying cause codes, showing what WHO rules of the implementation of IRIS, the CSO dual-coded the and instructions, it applied on a separate interface. 2015 mortality data to see the effects of the change in This was not the case with MMDS. software. Increased automation in coding means less manual intervention required by coders thus mitigating human misinterpretation and error. The international rules and What is IRIS? instructions for the selection of the underlying cause IRIS is a European Union-led software package initially of death leaves space for individual interpretation, using components of the MMDS software. IRIS is the and can result in a certain degree of variability of the software tool preferred by Eurostat and it is used by many tabulated underlying code among coders (Harteloh et of the EU Member States. It was developed by a core al., 2010). group of European countries. See link to IRIS website: See link: https://medwinpublishers.com/EIJ/ https://www.dimdi.de/dynamic/en/classifications/iris- EIJ16000102.pdf institute/index.html Bridging study When processing a death certificate, IRIS will first try to code all the diagnostic expressions on a death The CSO used IRIS to select cause of death codes for certificate (of which there may be several) and then the 2015 mortality records which were then compared select the underlying cause of death according to the to the original code assigned when using MMDS (the rules and guidelines published by the WHO in the ICD- coding done under both MMDS and IRIS were obviously 10 classification. See link to ICD-10 classification: http:// independent of each other). It is important to note that apps.who.int/classifications/icd10/browse/2016/en stillbirths and deaths identified as having external causes IRIS will try to automatically code as many death of death (suicides, homicides and accidents) continue to certificates as possible. Following this automated be coded manually by mortality coders. process, coders can edit the text of the certificate to There were 29,952 deaths registered in 2015 and each facilitate further processing or select the underlying was attributed an underlying cause of death code in cause of death manually. line with the WHO ICD-10 statistical classification of IRIS requires manual intervention by coders for diseases and related health problems. Having removed post-procedural disorders (E89, G97, H59, H95, unnatural deaths, a total of 27,502 mortality records were I97, K91, M96, N99), pregnancy, childbirth and the then processed in IRIS software, having previously been puerperium (O00-O99), accidental poisoning (X40-X49), processed using MMDS. The CSO used the 2017 IRIS complications of medical and surgical care (Y40-Y84) decision tables in the automatic coding process (55% of and sequelae of medical and surgical care (Y880-Y883). all records) and manually coded the remaining records (45%) using 2015 decision tables. Under MMDS the 2015 mortality data was coded using the 2011 decision tables Why move to IRIS? and the uncoded mortality records were coded using the 2014 decision tables (which were the most up to date The move to IRIS will lead to improvements in cause of version available at that time). death coding: IRIS is the most up-to-date software available for 1 Implementation of IRIS software for the automated coding of deaths in Ireland Comparing the MMDS and IRIS coding: 22,920 or 83.3% of records were attributed the exact 1,344 or 4.9% had a different UCOD code in IRIS same underlying cause of death code (UCOD) at four- The Comparability Ratio (CR) is defined as the number digit ICD-10 level of deaths coded under MMDS for a particular ICD code 24,509 or 89.1% had the same UCOD at three-digit level divided by the number coded under IRIS. Obviously the 25,262 or 91.8% had the same UCOD at two-digit level closer this ratio is to one then the less change there is in the total number of deaths for that ICD code under IRIS 26,158 or 95.1% had the same UCOD at one-digit level compared to MMDS. Table 1 Comparability Rate (MMDS V's IRIS) 2015 ICD_10 MMDS IRIS Comparability rate A 281 233 1.21 B 47 37 1.27 C 8,597 8,579 1.00 D 318 354 0.90 E 670 669 1.00 F 1,501 1,535 0.98 G 1,503 1,603 0.94 I 8,859 8,757 1.01 J 3,649 3,713 0.98 K 890 911 0.98 L 66 59 1.12 M 202 207 0.98 N 589 512 1.15 O 1 1 1.00 P 86 81 1.06 Q 155 171 0.91 R 88 80 1.10 Total 27,502 27,502 1.00 Per Table 1 above, the CRs show there is little or no Looking in more detail at the differences between the difference in the main causes of death groups i.e., ‘I’ two coding systems, Table 2 shows the changes between (diseases of the circulatory system), ‘C’ (neoplasms) and MMDS and IRIS at 1-digit level. For example there were ‘J’ (diseases of the respiratory system) which have a CR 96.3% of records given a ‘I’ code (i.e. diseases of the of 1.01, 1.00 and 0.98 respectively. circulatory system) both in MMDS and IRIS software The groups where the CR is further from one e.g., ‘A’ (certain while a further 1.7% of records that had been given an ‘I’ infectious and parasitic diseases), which has a comparability code in MMDS were attributed a ‘J’ code (i.e. diseases of rate of 1.21, are calculated on a much smaller number of the respiratory system) in IRIS. records (e.g. Group A accounts for 1% of death records). 2 Implementation of IRIS software for the automated coding of deaths in Ireland T R Q PO N M L K J I G F ED C B A MMDS otal 1.10.00.00.01.01.06.10.20.50.10.10.10.10.60.18.561.6A T 0.00.00.00.00.00.00.00.30.10.00.10.00.00.00.059.60.0Bable 2. Underlying cause of death b 2.31.30.00.01.41.00.01.10.70.20.00.20.15.798.62.11.4C 0.00.00.00.00.21.01.50.40.10.10.10.10.089.30.54.30.0D 0.00.00.00.01.22.50.00.60.20.20.20.192.10.30.06.40.0E 0.00.00.00.00.70.00.00.11.50.30.994.90.60.00.02.11.1F 2.31.92.30.00.72.50.00.32.00.396.11.90.30.30.10.01.8G 4.53.91.20.012.14.01.51.80.996.30.51.54.00.60.20.04.3Iy ICD-10 chapter MMDS(v 3.41.90.00.03.45.40.01.092.61.70.70.51.21.90.314.923.5J 1.11.30.00.00.70.51.593.10.50.20.10.20.60.00.12.13.9KIRIS 0.00.00.00.00.00.086.40.10.00.00.10.00.00.00.00.00.0L 0.00.00.00.00.880.73.00.10.50.10.20.10.10.00.00.01.1M 0.00.00.00.077.91.50.00.70.20.20.10.10.70.90.00.01.1Ner 100.0 tical) v Iris (horiz 0.00.00.00.00.00.00.00.00.00.00.00.00.00.00.00.0O 0.01.389.50.00.00.00.00.00.00.00.00.00.00.00.00.00.4P 0.088.47.00.00.00.00.00.00.10.10.90.30.00.30.00.00.0Q 85.2 0.0R ontal) 0.00.00.00.00.00.00.00.00.00.10.00.00.00.00.0 (%)T 100100100100100100100100100100100100100100100100100otal 27,5021555892028903,6498,8591,5031,5016703188,597281No. 88 861 66 47 3
no reviews yet
Please Login to review.