Introduction
Primary care delivery of hemodialysis for end-stage renal failure falls into three types within two general classes. The first class, Arterio–Venous access (AVA), can take the form of either the Arterio-Venous Fistula (AFV) or Arterio-Venous Graft (A-VG). The second class is the Permanent Venous Catheter (PVC), of which the better-known type is the Tesio line, a dual-tunneled catheter that improves blood flow during dialysis (National Kidney and Urologic Diseases Information Clearinghouse, 2008).
Each type bears advantages and drawbacks. By enlarging and strengthening the access vein, AFV’s promote blood flow, last longer without remedial surgery, and ensure fewer complications. However, the condition of the patient may not permit the waiting time of as long as 24 months for a proper fistula to develop. In which case, the alternatives are a vascular graft (or synthetic tube) that can be used within three weeks after placement or the single-lumen, twin-line Tesio-Cath that is immediately functional (Duncan et al., 2004).
Even after the vascular access has taken hold, all are subject to infection and constricted blood flow because of blood clotting in the access tubes. The trade-off for the improved blood flow of PVCs, for instance, is that they are the most susceptible to both problems. When that happens, medication to combat the infection and another procedure to remove or replace the catheter is called for.
Hence, both American and British guidelines recommend AVA as the gold standard while cautioning that no more than 10% of patients coming for hemodialysis (HD) to any Renal Unit should be allowed to opt for PVC. Beyond the high risk of sepsis, this recommendation is especially concerned with blood flow limitation below the optimal 1.2 for Kt/v or 65% for URR.
Despite this clear guideline on limiting the use of PVCs, the average incidence has been at least twice as much — 23% of HD patients in the USA and 28% in the UK.
Some of the reasons for this prevalence are well-known and others have not been disclosed (Pisoni et al., 2009)
Related to this “comfort level” with employing PVCs is the availability of the Tesio line type, a dual-tunneled catheter that has the advantage of maintaining desired flow adequacy. That being the case, Renal Units have been amenable to employing PVCs for patients with heart problems, to indulge the aesthetic desires of young women, or when the potential risks are justified by the fact that the patient is already on the waiting list for a kidney transplant.
True, AVA-type grafts are also subject to reduced flow adequacy because clotting can occur through neglect, poor hygiene practices in primary care, or, despite the best efforts, constriction of the access vein. In the latter situation, the AV graft may require angioplasty, a procedure to widen the small segment that has narrowed. Another option is to perform surgery on the AV graft and replace the narrow segment.
For their part, properly-implanted AV fistulas are the least susceptible to infection and flow inadequacy among all three access types.
Literature Review
The search strategy that ultimately proved more productive was that done on the Oxford Journals, Wiley InterScience, and National Institutes of Health databases employing, as a starting point, the keyword ‘hemodialysis’ and filtering first-line results with the key phrases ‘venous access’ and ‘catheter.’ The results are rather sparse, in part owing to the aforementioned emphasis on AVA in British and American peer-reviewed journals and the consequent relegation of PVC to second-class, risk-prone status.
The first of several related studies is a meta-analysis carried out by Murphy, White, and Nicholson (2000). The authors maintain that the continuing expansion of renal replacement under NHS auspices has increased the need for both the associated therapy programs and the methods of permanent vascular access that will suit more heterogeneous patient populations. If the prime objective is to ensure long-term patient survival, the authors maintain, the key ‘best practices’ turn out to be consultative vascular access planning, more consistent employment of preoperative imaging, and putting priority on grafting autogenous (the patient’s own) vein.
The fourth conclusion the authors advance is that renal units have come a long way in the detection or direct measurement of access flow. At present, it seems that surgical revision remains the best option for coping with venous stenosis (constriction or narrowing of the access vein), although percutaneous implantation of fistulas may offer some prospects.
The closest match to the proposed comparative study one finds in current literature is a prospective study done by Canaud, Leray-Moragues, Kerkeni, Bosc, and Martin (2002). In this instance, the study objective was to test for long-term dialysis adequacy of PVC versus AVA. To accomplish this, the research team embarked on a single-group, two-stage design of 42 patients on HD. All subjects were first implanted with ‘permanent’ dual silicone catheters (DualKT) for 12 months, followed by a further 12 months on AVA fistulas. Since there was no attempt to either control or manipulate any other treatment or care parameters, this design qualifies as a single group quasi-experiment where it would be logical to expect that any differences in dialysis performance can be ascribed chiefly to the access method.
As to critical outcomes, both methods surpassed recommended Kt/Vdp values ≥1.2, with AVA fistulas achieving 1.45 ±0.02 and PVCs 1.37 with a standard error of estimate at ±0.02. PVC was only slightly inferior to AVA in point of effective blood flow, venous pressure, and recirculation over time, albeit the authors conceded that the ‘slow blood flow technique’ used as the study instrument was prone to overestimating true access Rec.
Nonetheless, the research team may well have provided at least initial empirical evidence that PVCs are a viable alternative to AVA.
The generalizability of this prospective study was degraded by the fact that subjects were those available at the university hospital with which the authors were affiliated. This is convenience rather than random sampling. As well, it was not possible to run a parallel crossover design to test for sequence bias because the authors aver, the decision to commence HD is very often on an ‘emergency’ basis. Thirdly, the short duration of each treatment stage did not permit long-term assessment of flow adequacy, sepsis risk, patient survival, implant durability, or any of the critical dependent variables identified in section III-A., ‘Research Questions and Objectives’ (below), no matter that the authors would like to consider a year as ‘long-term exposure’ to the alternative treatments.
Conclusion
This admittedly cursory scan of the available literature in respect of hemodialysis has shown, first of all, that researchers are concerned about multiple issues such as electrolyte balance, vascular calcification, bone mineral density, the relative success of jugular versus femoral, and other implant sites, the common experience of thirst while on HD, survival and flow adequacy over really long periods (e.g. 35 years in a Slovenian retrospective study), vascular calcification, bone mineral density, even the ethics of surgeons in the United States implanting catheters as the first HD device in no less than 82% of patients embarking on HD. To date, however, there remains scope for a retrospective comparison of the two most popular methods with a sizeable sample and at least three years of ‘observational data.’
Research Proposal
Research Questions and Objectives
In general, this study will investigate the dialysis adequacy, durability and complication rates of AVA and PVC methods. At the very least, an update on the comparative benefits, risks and outcomes of the two access methods should serve to inform primary-care general practitioners, nephrologists and ultimately, ESRD patients themselves in the relevant local areas to make more prudent recommendations as to access devices.
In turn, the dependent variables shall be defined as follows:
(Based on Duncan et al., 2004)
Taking the cue from the state of knowledge revealed by the Literature Review and the operational variables under study (Depoy and Gitlin, 1994), the null hypotheses can therefore be articulated as follows:
- H01 : There is no difference between AVA and PVC in flow adequacy.
- H02 : There is no difference between AVA and PVC in average durability before replacement.
- H03 : There is no difference between AVA and PVC in mean complication prevalence.
Study Design
A retrospective cohort study is proposed, that is, one in which the medical records of end-stage renal failure patients already undergoing haemodialysis shall be accessed and analysed for meaningful differences in key outcomes by type of access: AVA and PVC (National Cancer Institute, n.d.). As with a prospective cohort study carried out by Chatterjee, Asari and Short (2004) on Tesio lines, this type of longitudinal research differs from case-control, “exposed versus non-exposed” designs in testing at least two patient groups subjected to different treatments but otherwise presumed to be the same in all relevant characteristics. Any difference in outcomes is therefore assumed to be due solely to the treatment rendered or in this case, access type implanted. Therefore, the complete nomenclature is a “retrospective comparative cohort study.”
Other important distinctions of this research design are that: a) the present study uses as a starting point patients already afflicted with end-stage renal failure and with HD access implanted whereas longitudinal and prospective studies routinely start with nonclinical populations (Bowling and Ebrahim, 2005); b) very little time need be devoted to data-gathering beyond retrieving copies of patient records from Renal Units and the National Kidney Care Audit, previously the ‘National Renal Audit’ (National Clinical Audit Support Programme, 2009).
Elaborating on the continued validity of retrospective cohort designs, already all of seven decades in the extant literature, Doll (2001) reminds us that the researcher merely collects data in the present time about latency, treatment exposure/replacement, and subsequent patient outcomes that have all already occurred in the past. To demonstrate the empirical rigour of retrospective comparative studies in respect of public health matters, the author compiled examples that included a 1933 study on contagion rates for tuberculosis where at least one family member had contracted the disease; odds of nickel refinery workers contracting lung cancer (for the period 1929 to 1938); the elevated odds of cancers of the lung and pleura amongst workers exposed to coal gas in the UK countryside up to 1938; the lifetime odds of contracting leukaemia and other fatal cancers amongst 8,500 survivors of the 1945 atomic bomb bombardment in Nagasaki and Hiroshima; and cancer incidence amongst ankylosing spondylitis patients administered varying schedules of radiotherapy. Clearly, retrospective cohort studies have made important contributions to advancing knowledge where cross-sectional research have limited application and longitudinal studies require enormous investments in time and finances to undertake.
By definition, this is a quantitative type of observational epidemiology study, given the expected sample sizes and the measurable nature of both the independent and dependent variables.
Since this study will be confined to a single metropolitan area, it is unlikely that the researcher will be able to control for, and create sub-samples according to, such key patient variables as age cohort, concomitant disorders such as diabetes, assisted in-home care, or mental disability.
Sampling Frame, Sample Size and Participant Selection
Also called ‘historic cohort study,’ this research shall assess the population of ESRD patients, resident in and around Brighton and Brighton and Hove, who had either AVA or PVC inserted between 1 January 2005 and 31 July 2008. The project plan aims to compile medical records for 200 patients implanted with each access type, net of incomplete records. Assuming a patient cohort weighted towards early January 2005, this means a maximum of 14,400 patient-months of follow-up (400 patients X a presumptive average of 36 months).
Systematic random samples shall be drawn, stratified within treatment centres (Doll, 2001) in the area such as Southmead Hospital, Bristol Royal Hospital for Children, and satellite dialysis centres yet to be sourced (National Health Service, 2009). At this time, the sole exclusion criteria shall be the implantation of fistulas since these are outside the scope of the proposed study objectives. Systematic random selection shall be achieved via interval sampling with replacement where interval = 200/population of patients with HD devices implanted from January 2005 to July 2008.
Data Selection and Collection
Other than the independent and dependent variables (see section A “Research Questions and Objectives” above), the study shall collect information on:
- Patient socio-demographics: paediatric or adult generally, by age cohort, gender, income class, body mass index, and ethnicity.
- Concomitant conditions and co-morbidities: length of pre-existent hypertension, juvenile or adult onset diabetes, cardiovascular disorders like congestive heart failure or AMI that necessitated surgery or ICU confinement, and other renal disorders.
- Treatment and diagnostic benchmarks: confinement frequency and mean length, urine output in millilitres per day, starting BUN, uric acid, and creatinine counts, and patient survival.
Data Analysis
Given that the data will be a combination of categorical and ratio variables, data analysis will cover descriptive statistics, tests for significance of differences, and inferential statistics. At the first stage, the composition and distribution of the categorical independent variables AVA and PCV will be examined employing descriptive statistics and plots: pie charts, and stem-and-leaf plots for replacement episodes. The second stage will comprise simple cross-tabulation of the two nominal variables against all ratio variables, as well as calculations of mean, median, standard deviation, minimum, maximum, quartile and percentile analysis as appropriate, supported by histograms, stem-and-leaf plots and box plots. In the third stage, tests of significance will be carried out employing, by way of example, ANOVA for incidence of catheter-related sepsis (CRS) across access device groups, the T test for prevalence and frequency of failed catheters across the categorical variables, and the Z test between diabetic and non-diabetic patients, as exemplified in a retrospective prevalence study of sepsis and patient survival when using tunnelled Tesio lines. Fourth, inferential statistics shall be employed to assess the conclusiveness of outcomes for the two ‘treatments’ being compared. For example, logistic regression will yield odds ratios that are the counterpart to relative risk measures employed in prospective cohort studies. For example, odds ratios were calculated in a retrospective study on long-term mortality after gastric bypass surgery. Step-wise multiple regression may also prove useful as a means of controlling for heterogeneous patient profiles or testing the effects of mediating variables (e.g., the vascular conditions such as diabetes and heart disease) common to elderly patients with ESRD (Chatterjee, Asari and Short, 2004; Altman, 2008; Department of Health, 2007; Adams et al., 2007).
Ethical Issues and Research Governance
The normal concerns and safeguards in respect of patient privacy and rights of consent will be obviated by requesting the two institutions in question for medical records stripped of names, addresses and contact numbers. And in any case, the study will report solely on summary measures, with no case histories that might enable readers to guess identities.
Whilst there is no requirement for voluntary participation, the researcher shall submit to the supervision of the (INSERT UNIVERSITY NAME) Institutional Review Board under the guidelines for undertaking a scientific investigation and being designed to contribute to generalisable knowledge. Further, the proponent undertakes to submit the full research protocol for review.
Time Scale
Limitations
Other than the evident one that this research is applicable principally to the ESRD patient population undergoing HD in Brighton and Hove, the ordinary limitations to retrospective studies apply:
- The absence of personal data about eating and drinking habits precludes analysis of dietary factors as mediating variables in understanding renal function outcomes.
- Masking of possibly important patient variables that are not routinely gathered from renal patients.
- Given the comparatively small sample size, this study shall not consider implantation size as a criterion independent variable.
- Mobility of patient families, especially outside the area of concern, may well underestimate adverse long-term outcomes, including death. The researcher trusts, however, that winnowing out of incomplete records will effectively remove such a source of bias. That is, a patient with no return record after a certain date can be presumed to have moved elsewhere rather than died.
References
Adams, T. D., Gress, R. E., Smith, S. C. et al. (2007). Long-term mortality after gastric bypass surgery. N. Engl. J. Med., 357 (8): 753–61.
Altman, D.G. (2008). Practical statistics for medical research. London: Chapman and Hall.
Ansell, D., Feest, T., Byrne, C. & Ahmad, A. (2003). UK Renal Registry: The sixth annual report. The Renal Association. Web.
Bowling, A. & Ebrahim, S. (eds.) (2005). Handbook of health research methods: Investigation, measurement and analysis. Milton Keynes: Open University Press.
Canaud, B., Leray-Moragues, H., Kerkeni, N., Bosc, J.-Y. & Martin, K. (2002) Effective flow performances and dialysis doses delivered with permanent catheters: A 24-month comparative study of permanent catheters versus arterio-venous vascular accesses. Nephrol Dial Transplant, 17: 1286-1292
Chatterjee, D., Asari, A. & Short, C. D. (2004). Tunnelled Tesio lines: A retrospective prevalence study of sepsis and line survival. Manchester Royal Infirmary. Web.
Department of Health (2007). The National Service Framework for renal services: Second progress report. Web.
DePoy, E. & Gitlin, L. (1994). Introduction to research: Multiple strategies for health and human services. Chicago: Mosby Yearbook.
Doll, Sir R. (2001). Cohort studies: History of the method II. Retrospective cohort studies. Soz.- Präventivmed, 46: 152–160.
Duncan, N. D. C., Singh, S., Cairns, T. D. H., Clark, M., El-Tayar, A., Griffith, M., Hakim, N., Hamady, M., McLean, A. G., Papalois, V., Palmer, A., & Taube, D. (2004). Tesio-Caths provide effective and safe long-term vascular access. Nephrology Dialysis Transplantation, 19 (11): 2816-2822.
Murphy, G. J., White, S. A., & Nicholson, M. L. (2000). Vascular access for haemodialysis. British Journal of Surgery, 87 (10): 1300 – 1315.
National Clinical Audit Support Programme (NCASP) (2009). National kidney care audit. The Health and Social Care Information Centre, NHS. Web.
National Health Service (2009). National kidney care audit, vascular audit: Renal unit participation status. The Health and Social Care Information Centre. Web.
The National Kidney and Urologic Diseases Information Clearinghouse (NKUDIC) (2008). Vascular access for hemodialysis. National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). Web.
National Cancer Institute (n.d.). Retrospective cohort study. U.S. National Institutes of Health. Web.
Pisoni, R.L., Arrington, C.J., Albert, J. M., Ethier, J., Kimata, N., Krishnan, M., Rayner, H. C., Saito, A., Sands, J., Saran, R., Gillespie, B., Wolfe, R. A., & Port, F. K. (2009) Facility hemodialysis vascular access use and mortality in countries participating in DOPPS: An instrumental variable analysis. (In press) Am J Kidney Dis. Web.