Intradialytic hypotension (IDH) is a common complication of haemodialysis (HD), the prevalence is 5–30%.1 This wide range is due to a lack of uniformity in definitions of hypotension used across different studies.2 It was recently demonstrated that nadir-based definitions are more consistently associated with mortality in predictive terms3 as compared to traditional definitions, such as those proposed by the National Kidney Foundation Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines4 or the European Renal Best Practice Guidelines.5 In addition to differences in prevalence, IDH usually presents high variability between patients,1,6 which makes the question of what definition of intradialytic hypotension to use even more difficult to answer.

The high variability of blood pressure resulting from HD makes it hard to stablish objective of blood pressure control.7 Blood pressure tends to be higher in predialytic periods, it decreases during the dialysis procedure and increases again in the next interdialytic period.2 This variability is more marked in patients with conventional regimens (three sessions per week), since those with more intensive dialysis regimens such as daily HD have better blood pressure control without requiring antihypertensive agents.7 There are factors dependent on the individual patient that may influence a predisposition to episodes of intradialysis hypotension, these include capacity of cardiac response during ultrafiltration (UF) and residual renal function (RRF).8

The objective of the study was to analyse the clinical and dialysis-related characteristics of different definitions of IDH and the relationship with mortality in a cohort of HD patients.

In a cohort of prevalent HD patients, we have demonstrated the power of IDH to predict non-fatal cardiovascular events and of overall mortality, especially with asymptomatic (nadir-based) definitions. This magnifies the importance of an episode of hypotension on dialysis regardless of whether or not it is accompanied by associated symptoms.

Suitable evaluation of the impact of IDH depends largely on the use of a definition that encompasses various clinical scenarios and approximates regular practice in a dialysis session as closely as possible. Some definitions are based on the need for implementing corrective measures such as reducing the UF rate and administering a saline solution, whereas other definitions are based on an absolute drop in SBP compared to predialysis SBP or are nadir-based definitions without the presence of symptoms.1,2 However, clinical guidelines use a combination of these factors, i.e. a drop in SBP accompanied by a clinical event plus the need for intervention.1,2 For example, the KDOQI guidelines and the European good practice guidelines define IDH as a drop in SBP greater than or equal to 20 mmHg, or a drop in mean BP of 10 mmHg, plus the presence of symptoms of hypotension (headache, cramps, chest pain, decreased visual acuity or any symptom that could be attributed to a situation of tissue hypoperfusion), requiring or not intervention measures to increase blood pressure or improve associated symptoms.3,4

This study used the eight definitions of IDH most commonly found in the literature, including those recommended by European clinical guidelines and KDOQI guidelines.1–5 Recently, Flythe et al. showed the predictive power for mortality of the Nadir90 definition in patients with a predialysis SBP around 130 mmHg,3 a figure for predialysis SBP similar to that of our study, with the same result for predictive power for mortality for the Nadir90 definition. However, with the Nadir100 definition we did not find a relationship with mortality; this could be explained by the mean predialysis SBP in our series, among other factors. The predictive capacity of Nadir100 was more consistent in patients with a predialysis SBP in excess of 140 mmHg.3

After evaluation of the relationship between IDH and non-fatal cardiovascular events, it was observed that the definition of hypotension that maintained the most consistent prediction in the adjusted models was the combination of Nadir90 and a drop in predialysis SBP by 30 mmHg (Fall30Nadir90). It should be noted that the survival analyses showed that other definitions also have a predictive power for fatal and non-fatal cardiovascular events, although the adjusted models did not show their relationship to be as consistent as the Nadir90 and Fall30Nadir90 definitions, respectively. The low predictive capacity of symptomatic IDH in our study was surprising and might have been conditioned by the low prevalence of the KDOQI and/or HEMO definition.

Our data reveal the high variability of IDH between patients, this is similar other reports on IDH.6 Regarding the definition of symptomatic IDH (KDOQI), 39% of study patients did not have any session with symptomatic hypotension. By contrast, at least one episode of symptomatic hypotension was recorded in 60% of patients. Something similar occurred with nadir-based definitions, although a drop by 20 mmHg in SBP (Fall20) was a constant phenomenon, to such a point that just one of the patients in the study did not experience any episodes of hypotension according to this definition. We defined IDH as fulfilment of the definition of hypotension in more than 25% of the sessions analysed. This cut-off point was similar to that proposed by other authors.3 If this cut-off point is considered, definitions of symptomatic IDH and/or IDH requiring some sort of intervention had a prevalence of 10.3% and 20.6%, respectively. By contrast, the prevalence of Nadir100 was 50% versus 27.9% for Nadir90. These differences regarding the prevalence of IDH seen in our data are similar to those reported in other studies with an IDH prevalence of 5–30%, and largely depend on the definition used.1,2

IDH is determined by patient-specific factors (advanced age, diabetes, more time on HD, low predialysis SBP, heart failure, autonomic dysfunction) but also by HD-specific factors.8 At present, there is enough clinical evidence to characterise HD as a risk factor from a haemodynamic point of view, especially in a vulnerable vascular tree.9,10 The cumulative negative effect of haemodynamic stress resulting from HD may harm organ function.9,10 The best example of this clinical scenario is a disease known as stunned myocardium.11 A common condition in HD,12 it entails a loss of contractile function at the time when greater cardiac output is required to compensate for a reduction in blood pressure.9–12 In addition there is the usual compromise in CKD of adaptive mechanisms in the microcirculation, which is aggravated in the context of a greater requirement for oxygen as occurs in left ventricular hypertrophy.8

In our study, the presence of heart disease was only associated with a drop in predialysis SBP by 30 mmHg (Fall30), although advanced age and more time on HD were factors associated with a higher frequency of episodes of hypotension according to nadir-based definitions. Diabetes mellitus increased the likelihood of experiencing a symptomatic episode of hypotension (KDOQI). The negative effect from a haemodynamic point of view acts as an expansive phenomenon, to such a point that all body organs can be critically affected. Notably, recent reports at the brain level by Polinder-Bos et al. and Findlay et al. demonstrated a significant drop in cerebral blood flow during an HD session,13,14 a specular image of what occurs at the cardiac level, which the authors called "stunned brain".14

A drop in blood pressure on HD is not usually accompanied by a positive effect on the mortality observed in the general population,7 and this study found that a lower predialysis SBP increased the likelihood of experiencing hypotension according to nadir-based definitions. It also found that a lower predialysis SBP was a predictive factor independent of mortality; this finding was similar to other authors.15–20 Although we lack sufficient evidence to determine the optimum range of blood pressure on HD, a growing body of data suggests that the relationship between morbidity and mortality and blood pressure follows a J- or U-shaped curve.15–20 Li et al., Klassen et al., Zager et al. and Bansal et al. obtained results similar to those presented in this study with the common denominator of a higher risk of mortality in a situation of "normotension" or of hypotension in patients on HD with different ranges of predialysis SBP predictive of mortality (e.g. < 120 mmHg, <130 mmHg, <110 mmHg).16–19 It is likely that the paradoxic effect of this relationship is explained by greater vulnerability from a haemodynamic point of view in patients with heart disease,20 and in our study a predialysis SBP above 140 mmHg was associated with a lower mortality during follow-up.

The dialysis regimen of the study patients consisted of three sessions per week with an optimal dialysis dose in the 1224 dialysis sessions analysed, consistent with that recommended by dialysis suitability guidelines.21 Interdialytic weight gain was less than 2 kg, a typical mean in our setting,22 which explains the UF rate of 7.4 ± 2.9 mL/h/kg. This rate is considerably lower than the cut-off points of 10, 12 and 13 mL/h/kg proposed by other authors as risk factors for cardiovascular morbidity and mortality.23,24 However, it is above 6 mL/h/kg, which other authors have identified as the cut-off point where the linear relationship between mortality and UF rate begins.25 An interesting aspect of our data is that we did not find any differences between the UF rate and the different definitions of hypotension.

Uniform criteria with regard to the ideal range for blood pressure on HD have yet to be established in the field of nephrology. The KDOQI guidelines, with a grade C recommendation, define a target predialysis blood pressure as <140/90 mmHg and a target postdialysis blood pressure as <130/80 mmHg.7 Based on information drawn from observational studies, this practice could expose patients to a higher rate of intradialytic hypotension, with a resulting negative effect on morbidity and mortality.15–20 Therefore, clinical trials that more clearly define treatment objectives for blood pressure in patients on HD are needed.

A key consideration related to haemodynamic tolerability on HD is the estimated dry weight. This concept has evolved as has dialysis itself, representing a real challenge for the nephrologist given the poor precision of a physical examination. To date, it remains a matter of debate.26,27 According to the most classic definition, dry weight represents the state of "normohydration" or euvolaemia after the dialysis session; thus it was determined that a patient reached their dry weight when on dropping from that level they experienced an episode of IDH.22,26–28 Bioimpedance spectroscopy is a tool that may aid in estimating dry weight,27 although calculation of dry weight is more an art and a matter of trial and error than a mathematical formula, or the result of the application of a technique.22 According to this line of argument, other authors propose that a "permissive state of hypervolaemia" would ensure perfusion of different organs as it maintains a state of slightly increased extracellular volume, which they define as "functional dry weight", and would be a reasonable option to prevent damage caused by states of hypoperfusion in the context of IDH.27 Our results reveal that we found no significant differences between the different definitions of hypotension. This was particularly striking in symptomatic definitions but also striking in nadir-based definitions. Other factors may account for these data (e.g. a low UF rate), but it is necessary to stress the role of the nephrologist in "correctly" estimating dry weight on HD, whether to achieve the objective of a state of "normovolaemia" or in cases in which it is preferable to achieve that "permissive state of hypervolaemia" explained by other authors as a protective mechanism against tissue damage on dialysis.

This study was not free from limitations. For example, its small sample size resulted in a weaker statistical power and impeded our ability to generalise our results. Other critical matters could be the retrospective analysis of the dialysis sessions — although we included a reasonable number of sessions, which gives an overall idea of the situation of the patient on dialysis — and the observational nature of the study, which precludes demonstration of a causal relationship between hypotension and cardiovascular events.

In conclusion, IDH, even asymptomatic, may have power to predict f mortality and non-fatal cardiovascular events in prevalent HD patients. This highlights the need to focus the efforts on preventing haemodynamic stress in order to minimise tissue damage and the resulting negative effect on morbidity and mortality.

This study has received no specific funding from public, private or non-profit organisations.

SC: study design, data collection, statistical analysis, literature review, manuscript drafting. AV: statistical analysis, manuscript review. NM: statistical analysis, manuscript review. LS: data collection, statistical analysis. SA: statistical analysis, manuscript review. JMLG: study design, manuscript review. JL: manuscript review.

The authors declare that they have no conflicts of interest.