The Effect of Gender, Age, and Geographical Location on the Incidence and Prevalence of Renal Replacement Therapy in Wales

Hugo C van Woerden; * Jane Wilkinson; Martin Heaven; Jason Merrifield

Disclosures

BMC Nephrology 

In This Article

Methods

A census of all patients in Wales on RRT was undertaken on 30 June 2004. This date was chosen, as it is the middle day of the year and population projections are calculated for that day of the year. A protocol was developed with input from renal physicians and circulated to previously agreed contacts in all renal centres in Wales and those in neighbouring areas in England used to treat Welsh patients. Data gathered included modality of RRT on that date, postcode, date of birth, gender and an indicator of whether the patient had first started RRT on or after 1 July 2003.

To comply with data protection legislation, no names or other personal identifiers were collected. The data were cleaned by comparing all postcodes with a database of all current and past postcodes in Wales and then removing patients with non-Welsh postcodes. Patients with missing or invalid postcodes or dates of birth were checked with the provider Trust. Data sets were combined in a secure Microsoft Sequel Server 2000 database. Duplicate entries were identified as those where the date of birth, gender and postcode were identical. The number of deceased patients in the collated data set was estimated to ensure that "ghost" patients would not unduly inflate our RRT rates.

The study was conducted in conformity with the requirements of the Declaration of Helsinki. Advice was sought from relevant experts to ensure compliance with Data Protection requirements. The South Wales Renal Managed Clinical Network determined that formal ethical approval was not required, as identifiable data was only made available to Health Solutions Wales staff, who have authorisation from the Patient Information Advisory Group (PIAG) to handle Patient Episode Database Wales (PEDW) data under Section 60 of the Health and Social Care Act 2001. This view was supported by Health Commission Wales, an executive agency of the Welsh Assembly Government, who commissioned the study.

To clean the data set further, possible entries representing deceased patients were identified by comparing date of birth, gender and postcode in our data set against the Welsh NHS Administrative Register (AR) held by Health Solution Wales. This comparison was made separately for patients treated by Welsh NHS Trusts and for one English Trust where there was concern that there might be a significant number of deceased patients. However, entries potentially representing deceased patients could not be removed from the dataset as the matching process could suggest, but not confirm, that a particular matching entry might represent a deceased patient. Our methodology was designed to estimate the potential size of this problem without having access to NHS numbers and to determine whether the numbers involved were sufficiently small as to be ignored in our calculation of RRT rates.

Denominator data were obtained from mid-year estimates of the population of Wales in 2004 (by Local Health Board) calculated by the Office of National Statistics for each one-year age band from the age of one to 89 years. For the one-year age bands between 90 and 99 years data from the 2001 census was used, as mid-year estimates are not available for ages over 89 years. No patients with an age of over 100 years were identified in the database and RRT rates for 100 years and over were consequently set to zero. Rates (pmp) were calculated for each one-year age band for the three modalities of RRT, i.e. haemodialysis, peritoneal dialysis and transplantation.

Moving averages of 3, 5, 7, 9, 11, 13 and 15 years were explored using an equal number of years above and below the age in question. An 11-year moving average was identified as providing the best balance between retention of definition of changing features of the graph, minimising lags in the peaks and troughs in the graph, and yet smoothing spikes in the data. Data for some graphs were truncated below 18 yrs and above 90 years because of numerators of fewer than 5 cases in some of the individual one-year age bands.

To indicate the probability of being on a given form of RRT at any given age, the proportion of patients on each of the three modalities of RRT was calculated for each one-year age band and converted into a percentage so that the three percentages came to 100%. An 11-year moving average was also applied to this data as described above. All data was analysed and graphed using Microsoft Excel™.

Postcodes were converted into a grid reference and recorded as an Easting and a Northing. Grid references were then mapped to Local Health Board (LHB) areas in Wales. This allowed the calculation of acceptance and prevalence rates by LHB. 95% confidence intervals were calculated for these rates using a SQL procedure which performs the confidence interval algorithm described by Newcombe[12] and Wilson.[13]

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.

processing....