Dialysing women and men – does it matter?
To date in my blog I haven’t focussed much on dialysis, but a recent news article about a new dialysis machine made me realise there are large gaps in my knowledge!
Generally textbooks are boring, unless that have lots of pictures, so I plunged right in with a paper published in the Clinical Kidney Journal in April of this year.
No particular logic was behind my choice of paper, but it has a catchy title (used above) and, scanning the abstract, I noticed the study included over 10,000 patients – thats a pretty big study! So here are the results of my Tuesday afternoon reading.
An observational study is simply a record of what happens, there are no interventions, the investigators report and analyse a situation or process. The downside is that results will describe associations but do not indicate cause and effect and there are many potential explanations. Nevertheless, data from 10,984 individual patients who underwent 1,999,648 dialysis sessions – well, thats information waiting to be discovered.
The study took place in Turkey, so my first question is what issues if any this raises for transferability – or, simply put, can the results be generalised across other populations. Most of my readers titleare from UK or US, and Turkey, in comparison, historically spends some 80% less on health care. However, following a change of government in 2003 a Health Transformation Program was instituted and together with a period of sustained economic growth, this has seen dramatic improvements across the board for healthcare. This included implementing social security contracts with private hospitals, which means the dialysis population in the study is probably a reasonable reflection of the population and in turn can be compared with equivalent situations in both US and UK, at least as far as the data used in the study. The study does not look at the pre-dialysis diagnosis or the cause of renal failure, but it is worth noting that type 2 diabetes and obesity are more prevalent in Turkey.
The study population was patients undergoing haemodialysis at Fresenius Medical Centres – there are 54 in Turkey. About 75% of the patients had been on dialysis for more than 3 months, and the paper analyses their data separately from those newer to dialysis, calling the two groups prevalent and incident respectively. Both groups had roughly equivalent numbers of male and female patients.
The study analysed data which included demographics, clinical, biochemical, therapeutic and prognostic indicators. The aim of the research was to identify any gender differences that might be important in affecting the outcome of dialysis. In the background information authors state that mortality for dialysis patients is unacceptably high and they quote a figure of 36% 5 year survival rate (Collins et al, 2014). That suggests that if you start with 100 patients on dialysis, in five years time only 36 will still be alive. This figure seems very low to me but I have not at this stage investigated.
An enormous quantity of data was gathered in this study and they utilised an array of statistical methods that had me trembling – only seasoned mathematicians would be excited by terms such as “Mann-Whitney U-test” and Greek formulae. After years of trying Sedgewick’s statistical puzzles at the back of the BMJ the only ones I really understand are Kaplan-Meier estimated survival curves.
This study makes extensive use of these survival curves to reach their conclusions. They are based on comparing two groups, one with the parameter under investigation and one without, using probability to estimate whether one group has better chance of survival compared with the other.
For example, in figure 1, the male patients (bottom line) have a slightly lower chance of still being alive after 3 years compared with the female patients. We have to take great care in interpreting these graphs, some apparent differences may not be statistically significant and the smaller the differences between the two groups the harder it is to draw conclusions.
However, the authors have distilled some significant associations, which I will expand upon, but in short:
Better outcomes (longer survival) were associated with:
- Access using an arteriovenous fistula (AVF)
- Higher than ideal BMI
- Being female
- Interdialytic weight gain of less than 5.7%
Poor outcomes (shorter survival) were associated with:
- Lower than ideal BMI in men
- Interdialytic weight gain of less than 1.4%
- Very low blood pressure
- Men with haemoglobin (Hb) out of normal range
- Low phosphorus in both men and women
- Low serum albumin
- Parathyroid dysfunction
Nothing is simple here though, so we need to look more closely.
The most common type of access is using an arteriovenous fistula, but this takes time and surgery to setup so it isn’t at all surprising that the incident patients (including those recently started on dialysis) would be less likely to have this mode of access. Instead one can have a central line or a tunnelled catheter access. It is reasonable to assume that the sicker patients will be using central line access since this is the default option in emergencies and so a poorer outcome might not be surprising.
In this study women were less likely to have an AVF for access. The authors explain that, due to higher oestrogen levels, women are more likely to have problems of clotting in an AVF, which might link with surgeon preference for alternative access methods in women. Other countries and studies have reported different figures regarding modes of vascular access so direct comparisons should be avoided.
What can be stated about access is that in this very large study, patients who had an AVF for vascular access did better than those who did not have a working AVF.
You may have been surprised at the statement that higher BMI was associated with better survival. I was. But before you open that packet of biscuits, lets look more closely. The BMI reference range for ideal was 21-25. In the study population BMI ranged from 20.2 to 33. So none of the patients were dramatically obese or extremely underweight. That said, 3-year survival for underweight women fell to around 57%, with figures for those over the reference range (over 25) reaching nearer 80%. For men the difference reached similar proportions.
It is likely this is not a simple association. For example, extremely ill patients needing acute dialysis are likely to be both underweight and more likely to die but obviously that does not mean that being underweight has caused the death. The authors use the term “reverse epidemiology” as a potential explanation of this issue. But the term is controversial, implying that epidemiology has different rules in dialysis patients compared with the general population, which of course is nonsensical. What it is demonstrating is that other factors, smaller differences, are probably influencing the outcomes and interfering with the generalisations. That does not diminish the strength of the observation, that in both men and women and both prevalent and incident patients, there was a better outcome for those above the ideal range of BMI.
The aim of the paper was to identify gender differences that might advise different management protocols for men and women. The first gender relevant issue to note was that women in the study tended to be older than men. This has been reported in other studies. It could mean that more older women get chronic kidney disease, or that women are older when they reach the need for dialysis. Neither can be deduced from the study data. Women were noted to be more likely to need treatment with erythrogenic stimulating agents (ESA) due to more of them being identified as anaemic. Women were also more often given vitamin D treatment. Neither, however, were linked with outcome difference.
There are many factors implemented in control of phosphorus and bone mineralisation in CKD. Previous studies report that better outcomes are usually associated with having phosphorus levels within the reference ranges. This will be affected by diet, phosphate binders, and parathyroid function. The results for this study are inconclusive in this area.
Interdialytic weight gain:
This appears to be a minefield of contradictions and I struggled to make sense of the study. It seems that the only conclusion from this data is that if the weight gain between dialysis sessions is less than 5.7% dry weight, then the patients will do better, or have better chances of survival. It has been reported elsewhere that achieving this target is very difficult and may only be possible in the early stages when there may be some residual kidney function. This study did not identify duration of dialysis sessions or concentration of sodium in the dialysate, so the observation has to stand alone without making any assumptions as to cause or effect.
It probably isn’t a surprise that very low blood pressure is associated with a poor outcome. A pre-dialysis systolic BP of less than 100mmHg was associated with less than 30% 3 year survival. Within the normal range for BP (pre-dialysis systolic 120 – 140mmHg) there is no significant difference in outcomes for either gender.
There is no doubt that this study contributes a wealth of information to the field. With data from over 10,000 patients and almost 2 million dialysis sessions, the findings should not be ignored. Despite the inherent difficulties of an observational study, the bottom line clearly states that there are differences between men and women on dialysis that warrant further attention.
It is however, slightly more difficult to apply this on an individual level.
For myself, if and when the time comes, I would hope to have AVF access, a BMI just above the “normal” range, and maybe a good dose of vitamin D on board.