Interpretability of Home Blood Pressure Measurements in Haemodialysis A Post Hoc Analysis of a Randomised Cross‐Over Study

ABSTRACT

Aim

We aimed to assess patterns of home BP in a maintenance haemodialysis cohort in line with consensus guidelines and determine the agreement with in‐centre BP.

Methods

A post hoc analysis of a pilot‐scale, randomised two‐period cross‐over study comparing self‐monitoring of BP over 4 weeks with usual care in 41 haemodialysis patients. Dialysis systolic BP (SBP) was compared with (i) home SBP averaged over 24 h, (ii) home SBP measurements on non‐dialysis days between 6 pm–12 am and 6 am–12 pm.

Results

Thirty‐three participants with a mean age of 50 ± 14 years provided sufficient blood pressure data. Post‐dialysis SBP moderately agreed with home SBP measurements (K = 0.65) when averaged over 2 weeks on non‐dialysis days. The limits of agreement and mean bias were minimally different between 2‐week averaged home SBP and post‐dialysis SBP (mean bias −4.44 mmHg, 95% CI for mean difference between methods −61.63 to 52.59 mmHg), versus 24 h averaged home SBP and post‐SBP (mean bias −2.32, limits of agreement −61.63 to 56.98 mmHg). Home SBP measurements were as variable [average real variability (16 ± 6)] as in‐centre pre‐dialysis SBP average real variability (14 ± 5) and post‐SBP average real variability (13 ± 5).

Conclusion

This study demonstrates the variability of BP measurement patterns if participants are not limited to measuring BP at a pre‐specified frequency. Further studies are needed to assess optimal methods of standardising home BP monitoring in dialysis patients and to evaluate home BP thresholds that can be used as targets in randomised controlled trials.

Trial Registration: www.clinicaltrials.gov. NCT03403491

In this analysis, home blood pressure was as variable as peri‐dialysis blood pressure measurements. Averaging BP over 2 weeks failed to improve the utility of home BP, challenging current ERA‐EDTA guidelines and suggesting that the optimal methods for assessing home BP in dialysis patients remain unknown.

1. Introduction

Hypertension is prevalent among haemodialysis patients [1]. Blood Pressure (BP), when measured outside of the dialysis unit, is linearly associated with all‐cause and cardiovascular mortality [1, 2]. The aetiology of hypertension in haemodialysis patients is multifactorial, with volume fluctuations occurring during the inter‐ and intra‐dialytic interval being an important contributor to blood pressure and blood pressure variability [3, 4]. Despite the risks associated with poorly controlled BP, optimal BP targets and methods of BP assessment in maintenance dialysis patients remain unknown [5].

The relationship between BP in haemodialysis patients and mortality is complex. Increased mortality is evident in both low and high quartiles of blood pressure as measured by peri‐dialytic BP measurements in large observational studies [1, 5]. The association between lower BP and mortality, however, may reflect confounding factors attributable to prevalent cardiovascular disease and frailty [1]. Smaller studies evaluating out‐of‐unit BP have demonstrated a trend of increasing mortality with increasing quartiles of BP [6]. Although adequately sized randomised controlled trials (RCTs) are lacking, two meta‐analyses have demonstrated a positive benefit of BP lowering on cardiovascular and all‐cause mortality in dialysis patients [1, 7, 8]. Observational studies examining the BP range associated with the lowest mortality rate are conflicting, meaning definitive evidence to support BP target recommendations does not exist [1, 6].

Evidence‐based consensus on the most convenient and accurate BP measurement technique that can be used to delineate prognosis and guide clinical management is additionally lacking. Standardised BP measurements, such as ambulatory BP, are important for determining prognosis and estimating treatment effects in research [9]. Clinically, BP measurements guide ultrafiltration goals and serve as actionable targets to reduce cardiovascular risk. Ambulatory BP monitoring is considered the gold standard in dialysis patients, but its use is limited by perceived treatment burden and cost [10]. Dialysis clinic BP measurements agree imprecisely with ambulatory BP monitoring [11], and provide inferior prognostic information [1, 2, 12]. In general, home BP is more reproducible [13], has a better agreement with ambulatory BP monitoring [14], and demonstrates a linear association with mortality risk [1, 15]. As ambulatory BP monitoring is considered cumbersome, home BP monitoring may be the preferred patient option [16]. In the absence of ambulatory BP readings, nephrologists in clinical practice are likely to attempt to base their BP clinical decisions on either in‐centre readings or home BP readings. Where ambulatory BP monitoring is unavailable, the European Renal Association‐European Dialysis and Transplant Association (ERA‐EDTA) guidelines suggest using averaged home BP, measured in the morning and the evening, on 6 non‐dialysis days over a 2‐week period in the diagnosis of hypertension [15]. Studies that assess the optimal standardisation of home BP measures in dialysis patients are needed to make sense of home BP patterns and pave the way for randomised trials [17].

We performed a post hoc analysis of a pilot‐scale, randomised two‐period cross‐over study (HOPE‐01) to assess patterns of home BP measurements in a maintenance haemodialysis cohort and their agreement with in‐centre BP readings.

2. Methods

2.1. Ethics

This study was approved by Beaumont Hospital Ethics Committee in October 2017. The study was conducted in accordance with the Declaration of Helsinki.

2.2. Study Design

A post hoc analysis of the Haemodialysis Outcomes & Patient Empowerment Study 01 (HOPE‐01) was conducted. HOPE‐01 was a pilot‐scale, prospective, randomised two‐period crossover study comparing self‐monitoring of weight and blood pressure via an electronic health journal (patientMPower platform) with usual care in 43 haemodialysis patients. HOPE‐01 consisted of a usual care run‐in period (2 weeks) followed by random order 2 × 4‐week crossover periods. The comparison was the active pMp app [+ digital weighing scales + BP monitor] for 4 weeks vs. a sham application. The primary outcome of HOPE‐01 was patient engagement with the platform to assess the feasibility of such an intervention in haemodialysis patients. A detailed description of the trial design. Data from the 4‐week intervention period was included in the post hoc analysis, as only this period contained home BP readings.

2.3. Study Population

Patients were recruited from two dialysis centres in Dublin, Ireland. Patients were included in the study if they were at least 18 years, had daily unrestricted access to a suitable smartphone or tablet device at home with an email address, had home broadband and/or mobile data as part of their mobile phone service, demonstrated understanding of correct use of the patientMpower app, digital weighing scales, BP monitor, and other study equipment, were capable and willing to perform measurements (e.g., weight, BP) at home and record information on the patientMpower app on a daily basis, and were willing to give informed consent. Patients were excluded if they had significant confusion or any concomitant medical condition which would limit their ability to record symptoms or other parameters. A study flow chart is available in the Supporting appendix, Figure 1.1.

2.4. Dialysis Clinic BP

Routine BP measurements were taken by dialysis nursing staff pre, intra, and post‐dialysis. Measurements were calculated using an appropriately sized blood pressure cuff attached to an oscillometric BP device integrated with the Fresenius CorDiax 5008 machine (Fresenius Medical Care, Bad Homburg, Germany).

2.5. Home BP

All patients were given an A&D model UA‐651BLE device for home BP measurements and provided with a device instruction leaflet on the correct BP measurement technique (Supporting file). The instruction leaflet follows the minimum guidelines for standardised BP measurement techniques, with the exception of providing advice relating to caffeine, smoking, and bladder fullness [18]. Compliance with measurement techniques or frequency was not assessed. If multiple BP measurements were taken within a 15 min window, the last 2 BP measurements were averaged for that time period. Patients who had > 2 home BP readings in a fortnight over the 4‐week intervention period were included in the analyses. Masked hypertension was defined as a 2‐week averaged pre‐dialysis SBP of < 140 mmHg and home SBP > 135 mmHg [19].

2.6. Statistical Methods

All analyses were performed using Python 3.6 on Anaconda Jupyter or R 1.4.1103. Categorical variables are expressed as counts and percentages, and continuous variables are expressed as mean ± standard deviation. Data were examined by looking at individual raw home BP measurements versus in‐centre systolic blood pressure (SBP). Data were subsequently analysed using 3 different categories: (i) pre‐ and post‐dialysis SBP was compared with home SBP averaged over 24 h; (ii) 2‐week averaged pre‐ and post‐dialysis SBP was compared with 2‐week averaged home BP data on dialysis and non‐dialysis days; (iii) 2‐week averaged dialysis clinic (SBP) was compared with home SBP measurements on non‐dialysis days between 6 pm–12 am and 6 am–12 pm. Two‐week averaged data in the am and pm was chosen to approximate ERA‐EDTA guidelines related to home BP monitoring. Home BP was used as the reference method for analysis. Agreement was determined using kappa statistics for BP categories. A cut‐off of > 140 mmHg was chosen as the definition of systolic hypertension in accordance with the International Society of Hypertension (ISH) guidelines [20]. In addition, repeated measures Bland–Altman plots, using the method proposed by Bland and Altman, were calculated to demonstrate the level of agreement between pre‐ and post‐dialysis systolic BP (SBP) and home BP [21]. Blood pressure variability was analysed using average real variability (ARV), coefficient of variation (CV), and standard deviation (SD) as variability metrics. To increase the number of data points, a moving window with a step size of 2 days and window size of 2 weeks was used to calculate the variability metrics. ARV was calculated using the averages of absolute differences between consecutive BP measurements, which also accounts for the order in which the measurements are obtained. A linear mixed‐effects model was used to examine the association of repeated measures of BP and BPV in individuals with ultrafiltration volume, adjusted for age, sex, time, and blood pressure medications. A random intercept with a random slope was chosen to allow the effect of ultrafiltration volume to vary between individuals. Comparisons between the low and high blood pressure variability groups were conducted using a paired t‐test. Hypothesis testing was based on a nominal 2‐tailed significance level of 0.05.

3. Results

3.1. Patient Characteristics

Out of 41 patients, 33 had sufficient home BP measurements for analysis, defined as > 2 BP measurements within a 2‐week period. Table 1 summarises the demographic characteristics of the patient population. Patients measured their home BP for a median of 20 (IQR = 8.5, 24.5) days over the 4‐week intervention period. There were 425 clinic pre‐ and post‐dialysis BP measurements and 401 home BP measurements. Of these measurements, 269 were taken on the same day as dialysis, and 404 measurements were taken on interdialytic days. Supporting Table 1 summarises the number of patients measuring their home BP within a 2‐week period. The mean pre‐dialysis SBP was 148 ± 13 mmHg, mean post‐dialysis SBP 133 ± 12 mmHg, and mean home SBP was 137 ± 14 mmHg.

TABLE 1.

Demographic characteristics of 33 study participants.

Characteristics	Overall
N	33
Demographics
Age (years)	50 ± 14
Sex (% men)	21 (64)
Race (% white)	32 (97)
Primary cause of ESKD
Glomerular	15 (45)
Diabetes	6 (18)
Hypertensive	1 (3)
Genetic	6 (18)
Other	5 (15)
Co‐morbidities
Heart failure	4 (12)
Diabetes mellitus	8 (24)
Hypertension	25 (76)
Peripheral vascular disease	7 (21)
Dialysis sessions per week	3 ± 0.25
Dialysis time (min) per session	236 ± 28
IDWG (kg)	2.13 ± 0.7
Ultrafiltration rate (mls/h)	673 ± 169
Ultrafiltration volume (mls)	2689 ± 768
Anti‐hypertensives
ACEi	6 (18)
Beta blocker	16 (48)
Calcium channel blocker	10 (30)
Diuretic	12 (36)
Alpha‐blocker	1 (3)

Note: Data are presented as number (%) and mean ± standard deviation unless otherwise specified.

Abbreviations: ACEi: Angiotensin converting enzyme inhibitor, ESKD: end‐stage kidney disease, IDWG: interdialytic weight gain, kg: kilograms, mins: minutes, mls: millitres.

3.2. Agreement With Home SBP

A summary of the findings between the 3 main comparison groups is available in Table 2. BP measurements were averaged over 24 h and compared with pre‐ and post‐dialysis SBP. Using this method, post‐dialysis SBP and pre‐dialysis SBP demonstrated a weak agreement with home SBP (K = 0.48 and K = 0.40, respectively) using a cut‐off of > 140 mmHg. The mean bias between post‐dialysis SBP and home SBP was −2.32 with wide limits of agreement (95% limits of agreement for mean difference between methods of −61.63 to 56.98 mmHg). Pre‐dialysis SBP and home SBP readings had a mean bias of 10.93 (95% CI for mean difference between methods −50.12 to 71.98 mmHg) [Figure 1].

TABLE 2.

Summary of Bland–Altman and kappa analyses for three comparison groups.

	Mean bias	95% limits of agreement	t‐Stat	df	p	kappa
Two‐week average: post‐dialysis SBP compared to home SBP on non‐dialysis days	−4.44	−61.28 to 52.39	−2.62	61	0.01	0.65
Two‐week average: pre‐dialysis SBP compared to averaged home SBP on non‐dialysis days	9.78	−46.79 to 66.35	4.8	61	< 0.00	0.45
Post‐dialysis SBP compared to 24 h average home SBP	−2.32	−61.63 to 56.98	−1.83	284	0.07	0.48
Pre‐dialysis SBP compared to 24 h average home SBP	10.93	−50.12 to 71.90	8.1	284	< 0.00	0.40
Two‐week averaged: post‐dialysis SBP compared to all home SBP	−2.76	−59.68 to 54.16	−1.62	64	0.12	0.65
Two‐week average: pre‐dialysis SBP compared to all home SBP measurements	10.93	−50.12 to 71.98	5.92	64	< 0.00	0.36

Abbreviations: df: degrees of freedom, kappa: kappa statistic, SBP: systolic blood pressure, t‐stat: t statistic.

FIGURE 1.

Bland–Altman plots comparing (a) post‐dialysis systolic BP and 24 h averaged home systolic BP and (b) pre‐dialysis systolic BP and 24 h averaged home systolic BP.

The agreement between dialysis SBP recordings and home SBP categories was improved by using 2‐week averages, using a cut‐off of > 140 mmHg. Post‐dialysis SBP measurements, measured on non‐dialysis days and averaged over 2 weeks, demonstrated a moderate agreement with home SBP measurements (K = 0.65). The agreement between pre‐dialysis SBP and home SBP remained weak (K = 0.45). However, the limits of agreement for all SBP measurements remained unacceptably wide when continuous measurements were compared using a repeated measures Bland–Altman plot. The mean bias between 2‐week averaged post‐dialysis SBP and home SBP measurements on non‐dialysis days was −4.44 mmHg (95% limits of agreement for mean difference between methods −61.28 to 52.39 mmHg) [Figure 2]. Similarly, there was a positive bias of 9.78 mmHg between 2‐week averaged pre‐dialysis SBP and home SBP on non‐dialysis days (mean bias 9.78 mmHg, 95% limits of agreement for mean difference between methods −46.79 to 66.35 mmHg) [Figure 2]. Minimal changes in agreement metrics were demonstrated when all home SBP measurements (non‐dialysis and dialysis days) were included rather than non‐dialysis days only (K = 0.65 between post‐dialysis BP and home SBP) [Figure 3].

FIGURE 2.

Bland–Altman plots comparing (a) two‐week averaged post‐dialysis SBP and home SBP on non‐dialysis days and (b) two‐week averaged pre‐dialysis and home SBP on non‐dialysis days.

FIGURE 3.

Bland–Altman plots comparing (a) two‐week averaged post‐dialysis SBP and all home SBP measurements and (b) two‐week averaged pre‐dialysis and all home SBP measurements.

3.3. Blood Pressure Variability

Home SBP measurements were highly variable [Figure 4]. Home SBP ARV (16 ± 6 mmHg) was as high as pre‐dialysis SBP ARV (14 ± 5 mmHg) and post‐dialysis SBP ARV (13 ± 5 mmHg) [Table 3]. 18 study participants had evidence of low home SBP ARV (ARV < 15 mmHg) and 15 participants had high ARV (ARV > 15 mmHg). There was a statistically significant trend toward greater interdialytic weight gain (IDWG) and ultrafiltration volume among study participants with high home SBP ARV (IDWG: 2.02 kg versus 2.27 kg, p = 0.005, UF volume: 2671 mls versus 2710 mls, p < 0.001). Similarly, patients with high pre‐dialysis SBP ARV had greater UF volume removal (2586 mls vs. 2869 mls, p < 0.001), but not IDWG (p = 0.92).

FIGURE 4.

Illustrative SBP profile of (a) participant with high SBP ARV, (b) patient with low SBP ARB demonstrating (i) all home BP data compared to in‐centre data and (ii) averaged home BP data on non‐dialysis days.

TABLE 3.

Average real variability, coefficient of variation and delta of pre‐dialysis, post‐dialysis and home BP.

	Mean BP (mmHg)	ARV	Co‐efficient of variation	Delta
Pre‐dialysis SBP	148 ± 13	14 ± 5	9 ± 4	44 ± 18
Pre‐dialysis DBP	76 ± 10	9 ± 3	12 ± 6	30 ± 13
Post‐dialysis SBP	133 ± 12	13 ± 5	9 ± 3	41 ± 16
Post‐dialysis DBP	72 ± 8	8 ± 4	11 ± 5	26 ± 11
Home SBP	137 ± 14	16 ± 6	10 ± 4	46 ± 16
Home DBP	77 ± 7	8 ± 4	10 ± 5	24 ± 12

Note: Data is presented as mean ± standard deviation. The unit for BP is mmHg.

Abbreviations: ARV: average real variability, BP: blood pressure, DBP, diastolic blood pressure, SBP, systolic blood pressure.

Multivariable analyses were performed examining the association of BP and blood pressure variability with ultrafiltration volume adjusted for age, sex, and BP medications. Ultrafiltration volume had no effect on home SBP [β = −0.001 (−0.005 to 0.003), p = 0.55], pre‐dialysis SBP [β = 0.003 (−0.001 to −0.008), p = 0.65], pre‐dialysis SBP ARV [β = −0.003 (95% CI −0.003 to 0.002), p = 0.65] or home SBP ARV [β = −0.002 (95% CI −0.007 to 0.002), p = 0.34].

3.4. Discordant Data

Masked hypertension was identified in 9 out of 33 (27%) participants, defined as elevated home SBP > 135 mmHg and pre‐dialysis SBP measurements < 140 mmHg. This encompassed 6% of all paired dialysis and interdialytic sessions. There were 5 out of 33 participants identified who had evidence of SBP < 90 mmHg during the interdialytic period despite a post‐dialysis SBP > 90 mmHg at the preceding dialysis session. This involved 2% of all dialysis sessions.

4. Discussion

Much remains unknown about the optimal method of diagnosing and monitoring hypertension in haemodialysis patients. Our study's findings were threefold: (1) Home BP measured on interdialytic days demonstrated as much variability as in‐centre measurements; and isolated measurements lacked interpretability; (2) 2‐week averaged home SBP measurements demonstrated moderate agreement with 2‐week averaged post‐dialysis BP when measured as categories; (3) home BP measurements enabled the detection of possible masked hypertension and hypotension.

4.1. Agreement and Validation

Ambulatory blood pressure monitoring (ABPM) during the interdialytic interval remains the gold standard for BP monitoring [10, 15]. In contrast with the U‐shaped relationship between pre‐dialysis SBP measurements and mortality [1, 18], ABPM displays a linear association with mortality and end‐organ damage risk [2, 6]. Clinically, given financial constraints and the high treatment burden experienced in this population, repeated ABPM measurements may be unsuitable in assessing long‐term BP control [15, 16]. The optimal substitute for ABPM in the diagnosis and management of BP is unclear. Peri‐dialytic BP measurements have traditionally been used to assess BP control, but agree poorly with 44 h ABPM [11, 22, 23, 24]. A meta‐analysis of peri‐dialysis BP measurements demonstrated that pre‐dialysis BP overestimates ABPM during the 48 h interdialytic interval (agreement limits 41.7 to –25.2 mmHg), while post‐dialysis SBP measurements were found to underestimate BP (agreement limits 33.1 to –36.3 mmHg) [11]. Standardised clinic BP measurements, intradialytic BP measurements, and interdialytic BP measurements have all been used to improve the accuracy of hypertension diagnosis [14, 23, 25].

Studies evaluating the use of home BP measurements in dialysis are limited. Current literature suggests that home BP is more reproducible and accurate than peri‐dialysis measurements [1, 12, 15] and is a better prognostic indicator of mortality, cardiovascular risk, and target organ damage [12, 26]. Consensus documents have recommended using home BP as a simpler, more cost‐effective option to ABPM [15]. To our knowledge, this is the first analysis to evaluate home BP in dialysis patients by attempting to follow consensus guidelines by assessing BP in participants on non‐dialysis days, between 6 and 12 am/pm, over a two‐week period [15]. This study deviates from consensus guidelines in that subjects were included in the analysis if they had > 2 days rather than 6 days of inter‐dialytic readings. In our study, isolated measurements of home BP lacked interpretability. Interpretability of the data was improved by using averaged data. However, in contrast to prevailing guidelines, averaging BP data over 2 weeks minimally improved the utility of home BP measurements compared to averaging data over 24 h. Additionally, agreement metrics were minimally altered by including home BP measurements on dialysis as well as non‐dialysis days, suggesting that optimal patterns for rationalising home BP measurements remain unclear.

A failure to obtain standardised blood pressure readings contributes to the difficulty in interpreting home BP measurements. In reality, home BP readings obtained from patients often reflect ‘routine’ rather than standardised measurements [27]. Our post hoc analysis reflects real‐world practice. Patients were provided with an instruction leaflet and a validated BP monitor, but their compliance with BP measurement technique was not assessed. In a study of 106 patients, real‐world adherence to standardised home BP measurements was poor despite detailed instructions on the frequency and standardisation of measurements. Patients were asked to measure 28 standardised BP readings over a 2‐week period using a memory‐equipped electronic device and written logbook. Only 23.6% of patients were fully adherent to the measurement schedule. Fictional data were present for 16.0% of patients. There were 360 unscheduled readings performed by 61 patients (57.5%), and 34% of patients missed readings [27]. In our analyses, the frequency of BP measurements varied across participants, rendering the interpretation of individual participants' BP difficult, emphasising that further research is needed to rationalise BP patterns in this population before incorporating home BP in guidelines.

Notably, 2‐week averaged post‐dialysis BP readings had a moderate agreement with averaged home BP readings when analysed as an SBP cut‐off > 140 mmHg. Compared to pre‐dialysis readings, post‐dialysis readings have narrower limits of agreement with ABPM and have greater predictive power for mortality risk [11, 28]. Mitra et al. found that standardised BP readings recorded 20 min post‐dialysis did not differ significantly from ABPM measurements [23]. Similarly, standardised post‐dialysis BP readings > 122 mmHg taken 5 min after dialysis had sufficient sensitivity (80%) and specificity (79.7%) to predict hypertension [24]. Both intradialytic BP and intradialytic plus pre/post‐dialysis readings can improve the precision and accuracy of dialysis clinic BP measurements [14, 25]. In the same study by Sarafidis et al., intradialytic and intradialytic plus pre/post‐dialysis BP readings had an AUC value of 0.85 for the diagnosis of 44 h SBP > 130 mmHg [14]. Clinically, standardised post‐dialysis BP measurements and/or intradialytic BP measurements may be an alternative where home, ABPM, and automated BP measurements are unavailable.

4.2. Blood Pressure Variability

Both short‐term and long‐term BPV are independent risk factors for cardiovascular and all‐cause mortality [4, 29], and are higher in dialysis compared to CKD patients [3]. In comparison to 44 h ABPM, home BP measurements can assess BP and BPV longitudinally, enabling calculations of both short‐term (24 h) and long‐term (visit–visit) BPV [3]. Certain analyses have suggested that BP, BP metrics, and BPV rise progressively during the interdialytic interval concurrent with excess fluid [30, 31], and decrease with ultrafiltration and dry weight attainment [4]. Our small analysis neither refutes nor supports these findings. High home systolic BPV was associated with higher IDWG and ultrafiltration volume, a finding that was not demonstrated by regression analysis. Neither IDWG nor ultrafiltration volume indicates whether patients achieved euvolaemia by the end of dialysis.

The rationale behind tracking interdialytic BP is to separate BP and BPV measurements from haemodynamic fluctuations occurring peri‐dialysis. Peri‐dialysis BP measurements are highly variable. In a cohort of 9849 HD patients, the variability of within‐person dialysis BP measurements was greater than between‐person BP measurements [32]. This variability has been attributed to white coat HTN or volume fluctuations associated with ultrafiltration [3]. However, in our study, home BP measurements on non‐dialysis days, over a 2‐week interval, demonstrated as much variability as in‐centre measurements suggesting that the causes of variability in blood pressure in dialysis patients lie outside of the dialysis unit. The SBP ARV of home BP measurements (16 ± 6 mmHg) was as high as pre‐dialysis (14 ± 5 mmHg) and post‐dialysis SBP ARV (13 ± 5 mmHg). Of note, the frequency of home BP measurements was not standardised. Measurements were taken at multiple time points throughout the non‐dialytic day, which may have contributed to the BPV seen. However, the coefficient of variation of home SBP in our cohort is similar to short‐term BPV as measured by 44 h ABPM (44‐h SBP co‐efficient of variation 12.07 ± 2.59%) [29], although the home SBP ARV calculation is higher.

To our knowledge, this is the first study comparing home BPV with dialysis clinic BPV. BP remained highly variable despite measurements being separated in time from plasma volume fluctuations occurring on dialysis days, underpinning the multifactorial aetiology of hypertension in these patients [18]. BP and BPV are sensitive to ultrafiltration; however, haemodynamic responses vary across patients. In the Dry‐weight reduction in hypertensive haemodialysis patients (DRIP) trial, the decline in BP following dry weight reduction was inconsistent. The ultrafiltration‐attributable reduction in BP varied among participants, with higher BP responses seen in black participants [33]. Recommendations that suggest volume control as an initial step in hypertension management fail to consider the heterogeneity of BP across fluid states [5, 15, 34]. Only 15%–48% of patients have detectable volume expansion with co‐existent hypertension [35, 36]. Dialysis patients have an increased prevalence of arterial stiffness and endothelial dysfunction, meaning that patients are susceptible to organ ischaemia with fluid removal [1]. In patients with normo or hypovolaemia, an approach of dry weight probing in response to hypertension might precipitate adverse events. An individualised approach to BP management is necessary in dialysis patients, taking into account a patient's comorbidities, frailty and objective volume state [5].

4.3. Discordant Data

Home BP has value in detecting masked hypertension and white‐coat hypertension. The prevalence of masked hypertension in our patient sample was 27%; lower than previous reports that used home BP as a reference [34]. Uncovering masked hypertension may be beneficial in identifying patients requiring additional treatment. In an analysis of 97 patients followed up for 1 year in the Blood Pressure in Dialysis (BID) study, increased left ventricular mass index was associated with patients who displayed higher home SBP measurements compared to pre‐dialysis SBP [34].

Similarly, detecting hypotensive episodes outside the dialysis unit might facilitate individualised treatment plans. There was likely a selection bias regarding the relatively young cohort included in this trial. Nevertheless, episodes of interdialytic hypotension < 90 mmHg were detected in 5 patients despite a preceding post‐dialysis session BP > 90 mmHg. This encompassed 2% of all dialysis sessions. Lack of standardisation of BP measurements may have contributed to the discordancy seen.

5. Strengths and Limitations

This was a post hoc analysis of a RCT that evaluated the acceptability of an app involving self‐monitored blood pressure and weight. There was a selection bias in terms of the patients who consented, reflected in the relatively low median age and the high adherence to home BP monitoring. Our study is limited by its sample size and any conclusions drawn must be interpreted with caution. Home BP measurement technique differed from consensus recommendations and was not performed using a standardised technique or at pre‐specified times, which may have contributed to systolic BPV. Although patients measured their BP for 16 ± 10 days on average during the 4‐week intervention period, these measurements were not evenly distributed throughout the 4 weeks. As this was a post hoc analysis, ABPM was not used. Hence, there was no gold standard comparator to calculate agreement for peri‐dialysis and home BP measurements, and nocturnal BP was not assessed. No association with ultrafiltration volume, rate or interdialytic weight gain was demonstrated by regression analysis; however, there were insufficient home BP measurements taken during the interdialytic interval to measure a linear change in BP concordant with volume change.

6. Conclusion

Our study demonstrates the variability of BP measurement patterns if participants are not limited to measuring BP at a pre‐specified frequency. In this analysis, isolated measurements of BP lacked interpretability. Providing dialysis patients in this cohort with home BP monitors resulted in high levels of home BP adoption; however, without specific instructions, the frequency of monitoring varied and thus the interpretability of home monitoring was limited. Averaging BP over 2 weeks failed to improve the utility of home BP compared to 24 h averaged data, challenging current ERA‐EDTA guidelines and suggesting that optimal methods for assessing home BP in dialysis patients remain unknown. Adequately powered trials evaluating optimal BP targets are needed to inform guidelines before accepting home BP as a substitute for standardised clinic or ABPM measurements in dialysis patients.

Large‐scale randomised studies aimed at identifying optimal home BP thresholds individualised across dialysis patient phenotype and volume categories are needed; these can subsequently be used as targets in studies that examine the effects of interventions on hard outcomes such as mortality and cardiovascular hospitalisations.

Author Contributions

Research idea and study design: Vicki Sandys, Emer O'Hare, Conall M. O'Seaghdha, Donal J. Sexton. Data acquisition: Amy Hudson. Data analysis/interpretation: Vicki Sandys, Lavleen Bhat, Emer O'Hare, Donal J. Sexton, Conall M. O'Seaghdha. Manuscript preparation: Vicki Sandys, Donal J. Sexton. Each author contributed important intellectual content during manuscript drafting and revision, accepts personal accountability for the author's own contributions, and agrees to ensure that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved.

Conflicts of Interest

Vicki Sandys, Lavleen Bhat and Emer O'Hare have no financial interests or conflicts of interest to declare. Conall O'Seaghdha and Donal Sexton are shareholders in patientMpower. All co‐authors have seen and agree with the contents of the manuscript. We certify that the submission is original work and is not under review at any other publication.

Supporting information

Data S1: Supporting Information.

Sandys V., Bhat L., O'Hare E., Hudson A., O'Seaghdha C. M., and Sexton D. J., “Interpretability of Home Blood Pressure Measurements in Haemodialysis A Post Hoc Analysis of a Randomised Cross‐Over Study,” Nephrology 30, no. 9 (2025): e70112, 10.1111/nep.70112.