Q J Med 2003; 96: 739-745
© 2003 Association of Physicians
Variation in dialysis patient mortality by Health Authority
From the 1Hope Hospital, Salford Royal Hospital NHS Trust, Salford and 2The West Pennine Health Authority, UK
Received 16 April 2003 and in revised form 14 July 2003
 |
Summary |
|---|
 Top Summary IntroductionMethodsResultsDiscussionReferences |
|---|
Background: Maintenance dialysis is a relatively low prevalence, highly specialized, and labour-intensive treatment, which is usually delivered at regional centres serving many different health authorities. It is unknown whether a patient’s health authority, in many ways an accident of birth, influences long-term dialysis outcomes.
Aim: To study survival patterns in patients starting maintenance dialysis therapy in the north-west of England between 1990 and 1999.
Design: Retrospective analysis.
Methods: We analysed data from quarterly returns submitted to the West Pennine Health Authority from 10 dialysis centres, including health authority, dialysis centre, age, gender, mode of dialysis therapy, postal code and diabetic status. Postal codes were used to compute the distance from residence to dialysis centre and Carstairs index.
Results: There were 2458 patients from 18 health authorities. Survival on dialysis therapy differed by health authority (p < 0.0001). Health authorities were then grouped into socioeconomic families, using The Office of National Statistics health authority classification system (ONS1). ONS1 profiles at inception of dialysis therapy were also associated with disparities in survival, with subjects from Urban and Rural health authorities having longer survival than those from Mining and Industrial, Mature or Prospering health authorities (p < 0.0001).
Discussion: Survival on dialysis varies significantly by health authority. The interface between highly specialized, centralized, medical services and the health authorities they serve may be a major outcome determinant.
|
Introduction |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Maintenance dialysis therapy may be one the best examples of a low-prevalence, highly specialized and expensive long-term treatment. It is not surprising, therefore, that dialysis therapy tends to be delivered at regional centres, serving large population bases. It has been known for several years that illness and economic wealth are inversely correlated in many populations.1 Typically, postal codes have been used to estimate average socio-economic wealth. Such an approach measures socio-economic status at a level close to that of an individual patient, and is unlikely to aid in the process of marrying dialysis service delivery to larger scale needs. The aim of this study was to determine whether survival on dialysis differed by health authority in the north-west of England, in the time period 1990 to 1999.
|
Methods |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Participating dialysis units return data on all maintenance dialysis patients to the West Pennine Health Authority quarterly. At each quarter, these returns enumerate whether a patient is new or existing, and whether patients are no longer using a region’s dialysis services, because of transplantation, death or having left the region. We studied the 2458 patients starting maintenance dialysis between 1 January 1990 and 31 December 1999. The percentages of completeness for the following parameters were: health authority 94%; dialysis centre 93.0%; age 99.9%; gender 93%; mode of dialysis therapy 98.3%; postal code 66.8%; and diabetic status 54.3%. Postal codes were used to compute the distance from residence to dialysis centre, using the Royal Automobile Club website,2 as well as the Carstairs index of individual socioeconomic deprivation.3
Initially, health authorities were analysed as individual units, using the most common category as reference category. Health authorities were then grouped using the ONS (Office National Statistics) 1 classification system, which was designed to encapsulate key census fields. Seventy-one initial variables were subjected to two sequential cluster analyses, leading ultimately to six ONS families, with titles chosen to represent the main characteristics of the category. Physical and environmental characteristics were not factors in determining these names. ‘Rural’ areas are characterized by high values for agricultural employment, low values for deprivation, many 45–64 year olds, low public transport use, multi-car households, large dwellings with low occupancy rates, and a small proportion of immigrants. ‘Prospering’ areas have the highest proportion from social class 1 and 2, typically well-qualified subjects, the highest value for owner occupancy, high car availability, low unemployment, the lowest value for deprivation indicators. Double-income households without children are typical. ‘Mature’ areas are characterized by an older population, areas where major developments have happened in the past, high values for employment in the finances and services sector, and above-average deprivation indicators. ‘Urban’ areas are diverse, with a high proportion of the population employed in manufacturing, a high proportion of immigrants, and higher than average deprivation indices. ‘Mining, manufacturing and industrial’ have a younger population, the highest proportions in social classes IV and V, high unemployment rates, low qualification levels, and above-average proportions of lone parents and dependants with lone carers.4
Mortality analysis was performed using Poisson regression analysis, with censoring at transplantation, final follow-up, or 31 December 1999. Three basic types of models were used. In the first type of model, no covariate adjustment was performed. In the second, parameters with > 90% data completeness were included, namely dialysis centre, year of dialysis inception, initial mode of dialysis therapy, age and gender. The third model was similar to the second, with the addition of variables exhibiting < 90% completeness, namely the distance from the dialysis unit, the Carstairs index, and diabetic status. Only 32.1% of the datasets were complete with regard to the variables included in this third model.
|
Results |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Patient characteristics at dialysis inception are shown in Table 1. The mean age of the population was 53.6 years, and 39.3% were female. The patients received dialysis therapy at 10 hospitals, representing 32.6%, 25.7%, 20.4%, 10.9%, 4.2%, 2.2%, 2.1%, 1.1%, 0.3% and 0.3% of the patient population, respectively. Continuous ambulatory peritoneal dialysis, in-centre haemodialysis, home haemodialysis and minimal care haemodialysis were used by 55.6%, 37.8%, 6.0% and 0.6%, respectively. Some 20.5% started dialysis in 1990–1991, 20.7% in 1992–93, 24.2% in 1994–95, 17.4% in 1996–97, and 17.3% in 1998–1999. The patients resided at a mean distance of 20.5 km from the dialysis centre, and had a mean Carstairs index of 1.9. Diabetes mellitus was recorded in 14.1%.
|
Eighteen health authorities were represented, treating, in descending order, 15.4%, 15.3%, 13.9%, 12.2%, 11.3%, 7.5%, 7.5%, 6.9%, 4.5%, 3.6%, 0.8%, 0.4%, 0.2%, 0.2%, 0.1%, 0.1%, and 0.0004% (1 patient) of the patient population, respectively. A comparison of the units is shown in Table 1. The eight health authorities with 20 or fewer patients are not shown, for clarity. There were statistically significant differences in the primary renal centre, the mode of dialysis therapy, the distance from the dialysis unit, the Carstairs index, and the proportion of diabetics between the health authorities. Table 2 shows a cross-tabulation of health authority by ONS1 classification. The areas served by six health authorities were classified as ‘Mining, manufacturing and industrial’, four as ‘Urban’, two as ‘Mature’, one as ‘Prospering’, and two as ‘Rural’.
|
The overall mortality rate was 12.2 per 100 patient-years of follow-up. Mortality rates differed by health authority with and without covariate adjustment, as shown in Table 3. Table 3 also shows a mortality comparison when health authorities were categorized using the ONS1 Classification. In unadjusted analyses, the explained variability was lower for ONS1 Classification than for health authority, with
2 values of 78.0 and 145.5, respectively. Mortality rates were lower in those from Urban and Rural health authorities. When adjustment was made for the dialysis centre, age, gender, mode of dialysis and year of dialysis inception, the explained variability was lower than when health authorities were considered in isolation, with 2 values of 59.4 and 120.5, respectively (adjustment 1). The ONS1 classification was not associated with mortality when additional adjustment was made for Carstairs Index, distance to the dialysis unit, and the presence or absence of diabetes mellitus (adjustment 2). Dialysis centre was associated with outcome in unadjusted and adjusted models; distance to the dialysis unit was associated with outcome in unadjusted analysis only, while the Carstairs index had no association with outcome.
|
|
Discussion |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
We observed large differences in survival according to health authority. These survival disparities were observed in a relatively small, densely populated area, in which a given maintenance dialysis unit served many health authorities. The disparities could not be explained by differences in age, the primary dialysis centre, the year of inception or mode of dialysis therapy. The disparities could not be accounted for by differences in the prevalence of diabetes, the geographic distance from the dialysis unit, or the Carstairs index (an index of socioeconomic deprivation at the level of an individual census area). Finally, although the survival differed by the ONS1 classification, a general method of grouping health authorities, the association with mortality was less obvious than when health authorities were considered as individual units. If our findings are valid, they suggest that, even in a relatively circumscribed area, outcome on dialysis may be partly dependent on a chance occurrence, the health authority into which a patient is born.
Marked geographical variations in mortality in the general population have been recognised for some time. The North–South divide, with an increased mortality rate in the north of England, has been known for a long time. Similarly, in the US, a recent report by Mansfield et al. noted that there was an association between premature mortality and living in rural areas.5 This effect can be seen at a more local level, with two studies showing a strong association between mortality and deprivation indices at regional and local authority level.6,7
Our patients from Rural health authorities had lower mortality rates, an association whose basis is unclear. The possibilities could include better health in the rural population, referral of healthier rural patients, and better treatment on renal replacement of patients from rural areas. In all honesty, we can only speculate as to why dialysis subjects from rural areas in this study had longer survivals. The work of Roderick and colleagues showed an inverse relationship between acceptance rates and distance from the dialysis unit8 In our study, distance from the dialysis unit was not associated with mortality rates, after comorbidity adjustment. Systematic selection of healthier dialysis patients, either on the referral side, or on the acceptance side could explain the findings. Overall, health authorities, at the individual level, exhibited more robust associations with outcome than those seen when aggregated into ONS1 categories.
This study has several limitations. It is a retrospective study, based on returns from dialysis units on maintenance dialysis patients. It did not collect comorbidity extensively. There is a real chance that patients with acute failure may be included in some returns and excluded from others. It is debatable, however, whether these limitations could fully explain the differences in survival by health authority. The study suggests a lack of homogeneity in some, or all, of the following elements: referral patterns, resource availability, and clinical practice. Observation of such disparities, without explanation, suggests another disparity, namely the lack of high-quality prospective clinical governance at local and regional levels. The advent of large regional and national registries, which prospectively collect key data elements can only help in this regard.
|
Footnotes |
|---|
Address correspondence to Dr R.N. Foley, Nephrology Analytical Services, 914 S Eighth Street, Suite D253, Minneapolis, MN 55404, USA. e-mail: RFoley{at}nephrology.org
|
References |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
1. Hinkle LE, Whitney LH, Lehman EW, Dunn J, Benjamin B, King R, et al. Occupation, education and coronary heart disease. Science 1968; 161:283–46.
2. http://route2.rac.co.uk/webroute212/Names.asp
3. Carstairs V, Morris R. Deprivation: explaining differences in mortality between Scotland and England and Wales. Br Med J 1989; 299:886–9.
4. Wallace M, Denham C. The ONS classification of local health authorities of Great Britain. HMSO, 1996.
5. Mansfield CJ, Wilson JL, Kobrinski EJ, Mitchell J. Premature mortality in the United States: the roles of geographic area, socioeconomic status, household type and availability of medical care. Am J Public Health 1999; 89:893–8.
6. Drever F, Whitehead M. Mortality in regions and local authority districts in the 1990s: exploring the relationship with deprivation. Popul Trends 1995; 82:19–26.
7. Langford IH, Bentham G. Regional variations in mortality rates in England and Wales: an analysis using multi-level modelling. Soc Sci Med 1996; 42:897–908.[CrossRef][Web of Science][Medline]
8. Roderick PJ, Raleigh VS, Hallam L, Mallick NP. The need and demand for renal replacement therapy in ethnic minorities in England. J Epidemiol Commun Health 1996; 50:334–9.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. A. Rodriguez, S. Sen, K. Mehta, S. Moody-Ayers, P. Bacchetti, and A. M. O'Hare Geography Matters: Relationships among Urban Residential Segregation, Dialysis Facilities, and Patient Outcomes Ann Intern Med, April 3, 2007; 146(7): 493 - 501. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||