This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Research Protocols, is properly cited. The complete bibliographic information, a link to the original publication on http://www.researchprotocols.org, as well as this copyright and license information must be included.
Kidney disease is a significant burden on health systems globally, with the rising prevalence of end stage kidney disease in Australia mirrored in many other countries. Approximately 25% of the Australian population lives in regional and rural areas and accessing complex tertiary services is challenging.
We aim to compare the burden and outcomes of chronic kidney disease and end stage kidney disease in rural and urban regions of New South Wales (Australia’s most populous state) using linked health data.
This is a retrospective cohort study and we have defined two cohorts: one with end stage kidney disease and one with chronic kidney disease. The end stage kidney disease cohort was defined using the Australia and New Zealand Dialysis and Transplant Registry, identifying all patients living in NSW receiving renal replacement therapy at any time between 01/07/2000 and 31/07/2010. The chronic kidney disease cohort used the NSW Admitted Patient Data Collection (APDC) to identify patients with a diagnostic code relating to chronic renal failure during any admission between 01/07/2000 and 31/07/2010. Both cohorts were linked to the NSW APDC, the Registry of Births, Deaths and Marriages, and the Central Cancer Registry allowing derivation of outcomes by categories of geographical remoteness.
To date, we have identified 10,505 patients with 2,384,218 records in the end stage kidney disease cohort and 159,033 patients with 1,599,770 records in the chronic kidney disease cohort.
This study will define the geographical distribution of end stage and chronic kidney disease and compare the health service utilization between rural and urban renal populations.
Kidney disease is a significant burden upon health systems globally. The rate of new end stage kidney disease (ESKD) cases in 2012 was 357 per million in the United States, 108 in the United Kingdom and 110 in Australia [
The cost of renal service provision in the United States in 2010 was US$47.5 billion [
New South Wales (NSW) is Australia’s most populous state and includes 32.3% of Australia’s population, with approximately 25% living in rural and remote areas. There is currently inadequate data regarding differences in growth in demand for renal replacement therapy (RRT) in rural versus urban areas in Australia [
The Australian Diabetes, Obesity and Lifestyle Survey (AUSDiab) estimated that approximately 16% of the Australian adult population has a marker indicating the presence of kidney damage [
Validation of administrative datasets with renal disease specific registries has been conducted in Australia and a high level of agreement between the two collections was found [
Our study hypotheses are that rural patients with ESKD and CKD have higher mortality, higher hospitalization rates, and longer lengths of stay, require more inter-hospital transfers and have higher rates of late referral for RRT compared to similar urban patients. We expect that in an Australian setting, rural patients with ESKD use home-based therapies more often than urban patients, despite evidence to the contrary in a North American setting. We also expect that rural patients with CKD or ESKD and at least one other comorbid condition (cardiovascular disease, diabetes or cancer) have a greater burden of disease defined as a higher mortality, higher hospitalization rates, longer lengths of stay and more requirements for inter-hospital transfer compared to similar urban CKD and ESKD patients.
This is a retrospective cohort study consisting of two cohorts (see
The ESKD cohort will be identified using the Australia and New Zealand Dialysis and Transplant Registry (ANZDATA). This registry was established in 1963 and maintains records of all patients with ESKD receiving chronic renal replacement therapy (dialysis or transplantation) in Australia and New Zealand. All patients residing in NSW at initiation of ESKD treatment between 1/7/2000 and 31/07/2010 will be included in the ESKD cohort.
The CKD cohort will be identified from within NSW Admitted Patient Data Collection (NSW APDC) by the Centre for Health Record Linkage (CHeReL) [
Both cohorts will be linked to NSW Admitted Patients Data Collection (NSW APDC), the NSW Registry of Births, Deaths and Marriages (NSW RBDM), and the NSW Central Cancer Registry (NSW CCR). The NSW APDC records all admissions to all NSW health care facilities, the NSW RBDM records all births, deaths and marriages within NSW and the NSW CCR records all new cancers in NSW residents.
Those that are under the age of 18 at the commencement of RRT or at the time of their first admission with a code for CKD will be excluded as well as those that do not normally reside in NSW. Residence will be assessed on the basis of postal code at the commencement of RRT or at the first admission with a code for CKD.
ICD-10AMa codes used to identify CKD cohort.
ICD 10 code | Description |
N18.1 | Chronic kidney disease stage 1 |
N18.2 | Chronic kidney disease stage 2 |
N18.3 | Chronic kidney disease stage 3 |
N18.4 | Chronic kidney disease stage 4 |
N18.5 | Chronic kidney disease stage 5 |
N18.8 | Other Chronic renal failure |
N18.9, N18.90, N18.91 | Chronic kidney disease unspecified |
N19 | Unspecified renal failure |
N 16.0-N16.8 | Renal tubulo-interstitial disorders in diseases classified elsewhere |
I12.0, I13.1, I13.2 | Hypertensive kidney disease with kidney failure |
E10.2, E11.2, E12.2, E13.2, E14.2 | Diabetes with kidney complication |
N00-N07 | Chronic nephritic syndrome, Nephrotic syndrome |
N11.0, N11.1, N11.8, N11.9, N12 | Chronic tubulo-interstitial nephritis |
N14.0-N14.4 | Drug and other tubular conditions such as analgesic nephropathy |
N25.0, N25.1, N25.8, N25.9, N26 | Impaired tubular function and unspecified contracted kidney |
N27.0, N27.1, N27.9 | Small contracted kidney |
N28.0, N28.1, N28.8, N28.9 | Other disorders of kidney not elsewhere specified |
N39.1 | Persistent proteinuria |
N39.2 | Orthostatic proteinuria |
B52.0 | Plasmodium with nephropathy |
D59.3 | Hemolytic uremic syndrome |
E85.3 | Secondary systemic amyloidosis |
Q60.0-Q60.6 | Renal agenesis |
Q61.3 | Polycystic kidney disease unspecified |
T 82.4 | Mechanical complication of vascular dialysis catheter |
T86.1 | Kidney transplant failure and rejection |
Z94.0 | Renal transplant |
aInternational classification for diseases 10 – Australian Modification
Data linkage process chart.
The exposure is rural residence defined using the Accessibility/Remoteness Index of Australia (ARIA) [
For the ESKD and the CKD cohorts, the following outcomes will be compared amongst the categories of remoteness: mortality (derived from fact and date of death via the NSW RBDM); hospitalizations (number of hospitalizations and location of hospitalization derived from NSW APDC); length of stay (using the hospitalization data provided by NSW APDC); inter-hospital transfers (calculated using the admission and discharge data gained from the NSW APDC). For the ESKD cohort, an additional outcome of rate of late referral to specialist care (identified via ANZDATA using the late referral flag, which measures those referred to nephrology care who subsequently start RRT within 3 months) and patterns of use of RRT (identified and compared using data on modalities of RRT used by patients from within ANZDATA) will also be compared between the categories of remoteness.
For both cohorts (ESKD and CKD), we will identify those with an additional diagnosis of cardiovascular disease (ICD10-AM Codes: I00-I52.8, I170 to I99), diabetes (ICD10-AM Codes: E10-14), or cancer (ICD10-AM Codes: C00-D48) and compare the outcomes of mortality, hospitalizations, lengths of stay and inter-hospital transfers as defined above.
Data linkage is probabilistic using demographic markers such as name, date of birth, gender, country of birth, medical record number (MRN), date of first RRT, postcode at first RRT, treating hospital and date of death to link patients identified by ANZDATA to the NSW APDC, NSW RBDM and the NSW CCR [
Data linkage will be performed using the services and processes of the CHeReL. CHeReL was established in 2006 with the aim of linking multiple sources of data and maintaining a record linkage system that protects data privacy and is jointly managed by the Cancer Institute NSW and the NSW Ministry of Health. Each data custodian provides information relating to individual persons to the CHeReL. This information consists solely of personally identifying information, plus an encrypted source record number (which is the link to the health dataset records). CHeReL uses the personally identifying information to link records for the same person across different datasets, and assigns a ‘person number’ to each of these groups of linked records (note that this ‘person number’ never leaves the CHeReL). CHeReL then develops a set of ‘project person numbers’ (PPN), which identifies all the records that correspond to a single person. CHeReL uses the
Once the required linkage has been completed with the groups of linked records identified and PPNs allocated, the CHeReL removes all identifiable information from the linked data sets and sends the data back to the respective data custodians. The data sent to the custodians contains their own encrypted source record numbers plus corresponding PPNs. The PPNs indicate which records correspond to a single individual so that the researchers can combine data from the different data custodians. Each data custodian then removes the source record numbers, and provides the researchers with the PPNs and the associated requested health data. This process ensures that the researchers are provided with deidentified data in which re-identification is effectively impossible.
We will separate the ESKD and CKD cohorts into categories of remoteness using the ARIA index as explained above [
This study has been granted ethical approval in January 2012 by the NSW Population & Health Services Research Ethics Committee along with approval from all data custodians. As no identifiable data will be provided to the investigators the risk to privacy of participants from the misuse of personal information used in the record linkage process is extremely small. This risk is further minimized by separating the processes of record linkage and data analysis. All data will be reported in aggregated form and no reports or presentations will identify any individual or organization.
The linkage keys, which allow linking of the relevant datasets, are destroyed 12 months following the supply of the data. After this time there will be no potential to reidentify the data. The data will be stored on secure servers for five years to enable the researchers to answer any queries arising from the publications as per ethical approval.
11,036 patients were identified by ANZDATA, of whom 10,827 patients also had records within NSW APDC. A further 322 patients either had missing postcodes or a non-NSW postcode leaving a total of 10,505 patients with 2,403,455 records in the ESKD cohort. Based on the ARIA categories, 85.46% of patients (8978/10,505) live in highly accessible areas; 11.77% (1236/10,505) in accessible areas; 1.84% (193/10,505) in moderately accessible areas; 0.66% (69/10,505) in remote areas and 0.28% (29/10,505) in very remote areas. For the purposes of analysis, patients living in accessible, moderately accessible, remote and very remote areas were combined as the rural cohort – 14.54% (1527/10,505).
The CKD cohort comprised of 164,236 patients. Exclusion of patients with missing or non-NSW postcodes resulted in 159,033 patients with 1,599,770,776 records in this cohort. Based on ARIA categories, 84.05% (133,667/159,033) live in highly accessible areas; 13.14% (20,904/159,033) in accessible areas; 2.05% (3260/159,033) in moderately accessible areas; 0.65% (1027/159,033) in remote areas and 0.11% (175/159,033) in very remote areas. For the purposes of analysis, patients living in accessible, moderately accessible, remote and very remote areas were combined as the rural cohort – 15.95% (25,366/159,033). The baseline characteristics of both cohorts are detailed in
Baseline characteristics of ESKD and CKD patients in New South Wales between 01/07/2000 and 31/07/2010.
|
ESKD (Urban) |
ESKD (Rural) |
|
CKD (Urban) |
CKD (Rural) |
|
Age (median & IQR) | 61 (48-72) | 61 (48-71) | .43 | 75.0 (62-83) | 74.0 (62-81.8) | <.001 |
Male (%) | 5246 (58.43%) | 888 (58.15%) | .84 | 69,142 (51.73%) | 13,392 (52.80%) | .002 |
Indigenous Australians (%) | 157 (1.75%) | 166 (10.87%) | <.001 | 1161 (0.87%)a | 1143 (4.55%)a | <.001 |
Comorbidities (%) |
|
|
|
|
|
|
Diabetes (From ANZDATA) | 2745 (30.57%)b | 454 (29.73%) b | .51 | NA | NA | NA |
(From NSW APDC) | 1536 (17.11%) c | 310 (20.30%)c | .002 | 43,072 (32.22%) | 8,019 (31.61%) | .06 |
Cardiovascular disease (From ANZDATA) | 3164 (35.24%)b | 592 (38.77%) b | .008 | NA | NA | NA |
(From NSW APDC) | 1237 (13.78%)c | 220 (14.41%)c | .51 | 46,137 (34.52%) | 8,355 (32.94%) | <.001 |
Peripheral vascular disease (From ANZDATA) | 2060 (22.94%)b | 440 (28.81%) b | <.001 | NA | NA | NA |
(From NSW APDC) | 3 (0%)c | 1 (0%)c | .55 | 107 (0.08%) | 9 (0%) | .02 |
Chronic lung disease (From ANZDATA) | 1273 (14%) b | 291 (19%) b | <.001 | NA | NA | NA |
(From NSW APDC) | 198 (2.2%)c | 41 (3%)c | .25 | 11,545 (8.6%) | 2,256 (8.9%) | .18 |
aRecorded for 157,792 (99.21%) patients.
bFor the ESKD cohort, these were derived from ANZDATA.
cDerived using ICD – 10 codes from the NSW Admitted Patient Data Collection. For the CKD cohort these were derived using ICD-10 codes for the index admission and in any admission prior to the index admission from within the NSW Admitted Patient Data Collection. The ICD – 10 codes were as per AIHW:
The mortality linkage identified a total of 96,313 records (88,020 patients) comprising 5463 records (5028 patients) in the ESKD cohort and 90,850 records (82,992 patients) in the CKD cohort. The linkage with the NSW CCR identified a total of 40,668 cancer records (36,110 patients) comprised of 1905 records (1693 patients) in the ESKD cohort and 38,763 records (34,417 patients) in the CKD cohort.
This research which has identified 11,036 ESKD patients and 164,236 patients in the CKD cohort will define the geographical distribution of CKD and ESKD as well as the demand for RRT in the NSW population. It will compare and contrast health service utilization between rural and urban populations with a view to informing the design and implementation of strategies to provide appropriate rural health care in the future. We will be able to delineate areas of higher incidence and prevalence and aid prediction of the need for future renal services. Given that 32.3% of the Australian population resides in NSW, this research has relevance for renal policy nationally.
The ANZDATA registry has made significant contributions to our understanding of kidney disease. This study expands the scope of ANZDATA and therefore will increase our insight into the drivers of mortality and poor outcomes in the kidney disease population. The ANZDATA registry however only records patients with ESKD that commence RRT and there has been no avenue previously for obtaining data on those with CKD/ESKD who are not receiving RRT except in the context of clinical trials. Our study allows us to comment on longitudinal outcomes in treated and untreated ESKD patients in a geographical context.
A notable limitation of our study however, is that the ascertainment of CKD relies purely on coding practices and coding intensity. Whilst there is no Australian data estimating prevalence of CKD in an admitted patient cohort, making it difficult to comment on the accuracy of coding for the CKD cohort, this dataset will be an important baseline for future research. Linkage with the ANZDATA registry for the ESKD cohort provides us the opportunity to report on validation of coding for the ESKD cohort as well as their comorbidities. Administrative health data, such as that used in this study, may represent a cheaper and effective alternative to performing large de novo longitudinal studies or maintaining large datasets. If so, it may also be a sustainable long-term option for measurement of disease burden and informing service delivery. A further strength is that because Australia has universal health coverage, our study includes all patients with kidney disease over a 10 year period that have had contact with private or public health care facilities in NSW.
This is a large retrospective Australian cohort study of patients with ESKD and CKD that uses the linkage of an existing renal registry and administrative datasets to compare the burden and outcomes of kidney disease in rural compared to urban settings. The results will enhance our understanding of the capability of administrative data in measuring kidney disease in Australia, compare the burden and outcomes in patients with kidney disease between rural and urban settings, and contribute to the design and development of renal health service provision in future years.
Australia and New Zealand Dialysis and Transplant Registry
Accessibility/Remoteness Index of Australia
Australian Diabetes, Obesity and Lifestyle Survey
Centre for Health Record Linkage
chronic kidney disease
date of birth
end stage kidney disease
Geographic Information System
Human Resources Ethics Committee
International Classification of Diseases 10 – Australian Modification
Medical Record Number
New South Wales
New South Wales Admitted Patient Data Collection
New South Wales Central Cancer Registry
New South Wales Registry of Births Marriages and Deaths
project person numbers
Renal Replacement Therapy
United States Renal Data Service
The authors wish to acknowledge the support of ANZDATA. Dr Sradha Kotwal is supported by a National Health and Medical Council scholarship and this project received a project grant from Kidney Health Australia (Grant number: PG0912).
None declared.