Novel equations for estimating intraperitoneal pressure among peritoneal dialysis patients

ABSTRACT

Background

Increased intraperitoneal pressure (IPP) is associated with abdominal wall complications and technical failure in peritoneal dialysis (PD). Since the standard measurement of IPP is limited due to its cumbersome procedures, we aimed to develop and validate equations for estimating IPP.

Methods

We performed a cross-sectional study with a total of 200 prevalent PD patients who were divided into development and validation datasets after random sampling matched by body mass index. The IPPs were measured using the Durand method, with whole-body and abdominal anthropometry indices collected. Equations with 2.0-L and 1.5-L fill volumes were generated by stepwise linear regression modelling. The bias, accuracy and precision of the estimated IPP (eIPP) with 2-L and 1.5-L fill volumes were compared with actual IPPs by the Durand method. The eIPP for the 2-L fill volume was also compared with other existing equations.

Results

Two new equations incorporating waist circumference and height from the decubitus plane to mid-axillary line were generated. The eIPPs exhibited small biases in relation to the Durand method , with median differences of −0.24 cmH₂O and −0.10 cmH₂O for 2 L and 1.5 L, respectively. The precisions evaluated by the standard deviation of the absolute value of the differences were 2.59 cmH₂O and 2.50 cmH₂O, respectively. The accuracies evaluated by the value of the percentage of estimates that differed by >20% for the eIPP were 26% for 2.0 L and 27% for 1.5 L. Better bias, precision and accuracy were observed for the eIPP equation compared with other existing equations for the 2.0-L fill volume.

Conclusions

We provided two new equations developed from abdominal anthropometry indices to accurately estimate the IPP in the PD population.

Graphical Abstract

Graphical Abstract.

INTRODUCTION

Peritoneal dialysis (PD) is a cost-saving modality for patients with end-stage renal disease (ESRD), providing benefits in cardiovascular stability, preserved renal function and lifestyle flexibility. Despite PD achieving similar short-term and long-term survival compared with haemodialysis (HD) [1–4], the long-term use of PD is still restricted by several complications. One of these is abdominal wall complications such as hernia, leak and hydrothorax, which are closely associated with elevated intraperitoneal pressure (IPP) of infused solution into the peritoneal cavity [5, 6] Studies have reported a varying incidence of abdominal complications of 4.8–37% [5, 7–10], resulting in temporary or permanent transfer to HD [9, 11–13]. In addition, increased IPP can significantly reduce the efficacy of PD, mainly by affecting ultrafiltration and clearance of solutes [14–16]. Considering its close association with clinical outcomes, we think it is crucial to routinely monitor the IPP in clinical practice.

It is well known that the Durand method [17] is the gold standard for IPP measurement. IPP is not routinely measured for PD patients due to the cumbersome procedure of the Durand method, which strongly relies on the patient's cooperation and trained staff. Several equations for the estimation of IPP have been developed, most of which include variables such as intraperitoneal fill volume (IPV) [18–22], body surface area (BSA) [20, 21] or body mass index (BMI) [18, 19, 23]. Abdominal cavity volume examined through computed tomography (CT) was recently recognized as a new predictor of IPP [18]. Notably, these equations were derived from relatively small-size datasets (17–81 subjects) [18, 20–26] or for specific PD populations [18–20], which might limit their generalizability. In addition, anthropometry measurements for the whole body such as BMI and BSA [18–21, 23] were applied to develop equations for IPP, but not considering any variables related to the size of the peritoneal cavity. This partly explains the unsatisfactory performances of these equations, as reflected by the interclass correlation coefficient or the R² of these equations [18–21].

Therefore, we aimed to develop and validate a new equation for establishing IPP in larger-size adult PD populations, considering both whole and local anthropometry measurements. We further compared the precision and accuracy of the newly derived equation with existing equations developed from the adult PD population.

MATERIALS AND METHODS

Study design and patients

This is a single-centre cross-sectional study. We recruited clinically stable PD patients according to inclusion and exclusion criteria. The inclusion criteria were age 18–80 years, PD duration ≥1 month, and intraperitoneal fill volume (IPV) ≥1.5 L. Patients were excluded if they had peritonitis, cardiovascular or cerebrovascular complications or other acute complications, abdominal surgery or trauma within 1 month and/or a history of hernias, hydrothorax or subcutaneous or pericatheter leaks. The Ethics Committee of Peking University First Hospital approved the study protocol and we adhered to the Declaration of Helsinki. Each patient provided written informed consent to participate in the study. This trial is registered at ClinicalTrials.gov (NCT05313841).

Demographic, comorbidity and dialysis data

Demographic and clinical data included age, gender, dialysis duration, primary renal disease and age-adjusted Charlson comorbidity index [27]. Dialysis adequacy and residual renal function (RRF) were measured by collecting 24-hour urine and dialysate. Dialysis adequacy was measured as total urea clearance (Kt/V) and total creatinine clearance (CrCl). RRF was estimated using the average renal clearance of urea and creatinine. Weekly averaged urine and ultrafiltration were calculated from a clinical assessment questionnaire 1 week prior to the clinic day. The peritoneal solute transporter state was measured with the dialysis adequacy and transport test [28], which was calculated as the ratio of 24-hour dialysate creatinine to intraday blood creatinine.

Anthropometry data

Anthropometry data included height, weight, BMI (kg/m²), BSA (m²), waist circumference, the iliac bi-spinous diameter, distance from the symphysis pubis to the xiphoid process and height from the decubitus plane to the mid-axillary line. All the baseline anthropometry data were collected with an empty peritoneal cavity. Standing height was measured using a fixed stadiometer and weight was measured using a calibrated digital scale. BMI was defined as weight/height². BSA was calculated by the Du Bois–Du Bois formula (0.007184*weight^0.425*height^0.725) [29]. Other anthropometric parameters were taken in the supine position and measurements were done directly on the skin. Waist circumference was measured with a tape measure wrapped above the iliac crests around the level of the umbilicus [30]. The iliac bi-spinous diameter was measured between the outer margin of the anterior superior iliac spine [31]. The distance from the symphysis pubis to the xiphoid process was measured [32]. The distance from the mid-axillary line to the decubitus plane was also recorded.

Biochemical measurements

Blood samples were taken following an overnight fast. Biochemistry data in relation to haemoglobin, serum albumin, lipids, glucose, uric acid, urea, creatinine, calcium and phosphate were obtained using an automatic chemistry analyser (Hitachi Chemicals, Tokyo, Japan). Serum levels of high-sensitivity C-reactive protein (hs-CRP) were measured using immune rate nephelometry and normal values were <3 mg/L. Serum intact parathyroid hormone (iPTH) was measured using a chemiluminescence assay (reference range 15–65 pg/ml).

IPP

The actual IPP (aIPP) was measured as previously described by Durand et al. [17]. The peritoneal cavity was completely emptied before the test. The patient was placed briefly in a supine position on a horizontal plane and the height of the horizontal plane was standardized for all study subjects. A graduated column (cm) was bound to the abdominal cavity through the PD catheter, with the zero level placed on the mid-axillary line. A total of 2 L of lactate-buffered glucose dialysate (1.5% or 2.5%; Baxter Healthcare, Guangzhou, China) was infused into the peritoneal cavity. We then read the distance between the column of dialysate inside the PD catheter and the mid-axillary line in centimetres of water (cmH₂O) at non-deep expiration (IPP_exp) and non-deep inspiration (IPP_insp), calculating the mean of IPP_exp and IPP_insp. After outflowing 500 ml of the dialysate solution, we measured the aIPP again utilizing the same procedure. As a result, the aIPP was measured twice for each participant. All the measurements were performed by two skilled nurses, who had good interrater agreement.

The IPP was estimated using the newly derived equations and other existing equations developed from adult PD patients: the Castellanos equation [23], the Scanziani equation [19] and the de Jesus Ventura equation [21].

Statistical analysis

Normally distributed data are presented as mean ± standard deviation (SD). Non-parametric data are presented as median and interquartile range (IQR). Categorical variables are expressed as percentages or ratios. Student's t, non-parametric or chi-squared tests were used to compare the differences in variables between cohorts, as appropriate.

All participants were randomly stratified into the development dataset (50% of patients) and the validation dataset (50% of patients) according to their BMI levels (<22.4, 22.4–25.8 and >25.8 kg/m²). After applying univariate Spearman's correlation analysis to determine the relationship between variables (all demographic and biochemical measurements) and the aIPP, a stepwise linear regression analysis was performed to select potential predictors to incorporate into the eIPP in the development dataset.

To validate the performance of the new equation, the eIPP value was compared with the aIPP in the validation dataset using block analysis of variance, chi-squared tests and independent t-tests in the subgroups. Bias was assessed as the median of the difference between the estimated and measured IPP values, precision was assessed as the SD for the absolute value of the differences and accuracy was assessed as the percentage of estimates that differed by >20% from the measured IPP (1 − P20). Confidence intervals (CIs) were calculated using bootstrap methods (2000 bootstraps). Patients were categorized into subgroups according to gender (male and female) and age levels (<65 years versus ≥65 years) to examine the performance of the newly derived equation.

To compare the performances of the eIPP equation for a 2.0-L fill volume and several other existing equations, the bias, precision and accuracy of the Castellanos equation [23], the Scanziani equation [19] and the De Jesus Ventura equation [21] in reference to the aIPP were also calculated.

All probabilities were two-tailed and the level of significance was set at .05. Statistical analysis was performed using SPSS for Windows version 22.0 (IBM, Armonk, NY, USA). The research was in compliance with the Declaration of Helsinki and approved by the Ethics Committee of Peking University First Hospital [2011(357)]. Written informed consent was obtained from each patient.

RESULTS

Participant characteristics

A total of 200 PD patients completed the examinations and measurements required for this study. The measurements of aIPP were taken for 2.0-L and 1.5-L fill volumes. The mean age of our participants was 51.8 ± 12.9 years, 37.5% were female and 32.0% had diabetes. The average BMI and BSA was 24.2 ± 3.8 kg/m² and 1.8 ± 0.2 m², respectively. The whole dataset was randomly divided into a development dataset (n = 100) and a validation dataset (n = 100) (Fig. 1). The basic demographic and clinical characteristics of the development and validation cohorts are shown in Table 1. The proportion of diabetic patients in the validation cohort (39%) was higher than the proportion in the development cohort (25%, P = .034). Patients in the validation cohort had a slightly higher serum sodium and lower Kt/V levels than patients in the development cohort (P < .05). There were no differences in demographic, anthropometric, primary kidney disease, comorbidity and biochemical data between the groups (P > .05).

Figure 1:

Flow chart of the study.

Table 1:

Demographic and clinical characteristics of PD patients in the development and validation datasets.

	Cross-sectional datasets
Variates	Total (N = 200)	Development cohort (n = 100)	Validation cohort (n = 100)	P-value
Age (years), mean ± SD	51.3 ± 12.9	50.5 ± 13.0	52.1 ± 12.8	.399
Male, n (%)	125 (62.5)	57 (57.0)	68 (68.0)	.109
Diabetes, n (%)	64 (32.0)	25 (25.0)	39 (39.0)	.034*
Dialysis duration (months)	25.0 (10.0,54.8)	29.0 (12.3,55.0)	21.0 (7.0,53.3)	.854
Primary kidney disease, n (%)				.832
Glomerulonephritis	94 (47.0)	50 (50.0)	44 (44.0)	.398
Diabetic kidney disease	48 (24.0)	18 (18.0)	30 (30.0)	.047*
Hypertensive nephropathy	22 (11.0)	15 (15.0)	7 (7.0)	.071
Interstitial nephritis	8 (4.0)	4 (4.0)	4 (4.0)	1.000
Polycystic kidney	5 (2.5)	1 (1.0)	4 (4.0)	.176
Other causes	5 (2.5)	2 (2.0)	3 (3.0)	.653
Uraemia unknown	18 (9.0)	10 (10.0)	8 (8.0)	.623
Height (m), mean ± SD	1.67 ± 0.08	1.66 ± 0.09	1.68 ± 0.08	.131
Weight (kg), mean ± SD	67.9 ± 13.6	67.7 ± 13.6	68.0 ± 13.6	.895
BMI (kg/m2), mean ± SD	24.2 ± 3.8	24.4 ± 3.7	24.0 ± 3.8	.418
BSA (m2), mean ± SD	1.8 ± 0.2	1.8 ± 0.2	1.8 ± 0.2	.640
Waist (cm), mean ± SD	86.6 ± 10.9	86.8 ± 10.3	86.5 ± 11.5	.844
Iliac bi-spinous diameter (cm), mean ± SD	26.4 ± 2.0	26.5 ± 2.1	26.3 ± 2.0	.628
Distance from symphysis pubis to xiphoid process (cm), mean ± SD	32.9 ± 3.2	32.9 ± 3.0	33.0 ± 3.5	.870
Distance from mid-axillary line to decubitus plane (cm), mean ± SD	10.1 ± 2.7	10.3 ± 2.8	10.0 ± 2.7	.443
CCI score	4.0 (2.0–5.0)	4.0 (2.0–5.0)	4.0 (3.0–6.0)	.099
Prescribed IPV (L), mean ± SD	1.9 ± 0.2	2.0 ± 0.1	1.9 ± 0.2	.225
Ultrafiltration (ml)	393.9 (84.9–776.9)	407.4 (59.1–816.5)	386.5 (122.9–755.8)	.989
Urine volume (ml)	493.0 (0–1073.2)	485.8 (0–1103.0)	548.6 (0–1050.5)	.903
Total Kt/V, mean ± SD	1.9 ± 0.4	2.0 ± 0.4	1.8 ± 0.3	.022*
Total CrCl (ml/min/1.73 m2), mean ± SD	48.4 ± 10.8	48.9 ± 10.1	47.8 ± 11.6	.448
Renal Kt/V	0.3 (0–0.7)	0.2 (0–0.7)	0.3 (0–0.7)	.941
Renal CrCl (ml/min/1.73 m2)	4.5 (0–13.3)	4.5 (0–12.3)	4.7 (0–14.5)	.504
D/Pcr, mean ± SD	0.7 ± 0.2	0.7 ± 0.1	0.7 ± 0.2	.158
aIPP with 2.0-L fill volume (cmH2O), mean ± SD	16.3 ± 4.2	16.1 ± 4.2	16.5 ± 4.3	.484
aIPP with 1.5-L fill volume (cmH2O), mean ± SD	14.4 ± 4.1	14.3 ± 3.9	14.6 ± 4.2	.574
Laboratory data
Serum albumin (g/L), mean ± SD	37.5 ± 4.1	37.6 ± 4.2	37.4 ± 3.9	.714
Haemoglobin (g/L), mean ± SD	111.7 ± 13.2	111.3 ± 12.7	112.2 ± 13.8	.625
hs-CRP (mg/L)	2.1 (0.8–5.0)	2.1(0.8–6.3)	2.1 (0.9–4.6)	.191
Blood glucose (mmol/L), mean ± SD	5.8 ± 1.7	5.8 ± 1.5	5.9 ± 1.9	.678
Urea nitrogen (mmol/L)	24.2 (20.8–28.0)	24.5 (20.8–28.1)	23.3 (20.3–28.6)	.253
Serum creatinine (μmol/L), mean ± SD	975.2 ± 280.9	1004.3 ± 293.3	946.1 ± 266.2	.143
Serum calcium (mmol/L)	2.3 (2.2–2.4)	2.3 (2.2–2.4)	2.3 (2.2–2.4)	.351
Serum phosphorus (mmol/L), mean ± SD	1.8 ± 0.3	1.8 ± 0.3	1.8 ± 0.3	.639
Serum sodium (mmol/L), mean ± SD	139.0 ± 2.3	138.7 ± 2.0	139.3 ± 2.5	.041*
Serum potassium (mmol/L), mean ± SD	4.3 ± 0.5	4.3 ± 0.5	4.4 ± 0.5	.462
Total cholesterol (mmol/L), mean ± SD	3.9 ± 0.9	4.0 ± 0.9	3.9 ± 0.8	.834
Triglycerides (mmol/L)	1.4 (1.0–2.3)	1.5 (1.0–2.3)	1.4 (1.0–2.1)	.120
iPTH (pg/ml)	263.6 (169.7–389.3)	283.8 (178.6–393.2)	251.9 (164.9–385.0)	.450

Data are presented as median values with their lower and upper quartiles unless state otherwise. Statistical test: independent sample t-test for continuous variables and chi-squared test for categorical variables. *P < .05, comparison between development and validation datasets.

Development of new equations for estimating IPP

We constructed an eIPP equation using the development dataset. Spearman's correlation analyses showed that aIPP with a 2.0-L fill volume is significantly correlated with BMI, waist circumference, iliac bi-spinous diameter, distance from the mid-axillary line to the decubitus plane, diabetes, hs-CRP and triglycerides (P < .05) (Supplementary Table 1A). When the fill volume was 1.5 L, the aIPP was additionally correlated with weight and distance from the symphysis pubis to the xiphoid process (Supplementary Table 1B). No significant associations were found between the aIPP and height, BSA and other biochemical data. All variables selected from the Spearman's correlation analyses were included to examine independent effects by a multivariate regression analysis with stepwise procedure. As Table 2 shows, waist circumference and distance from the decubitus plane to mid-axillary line constructed the best-fit regression model. The R² values for the newly derived eIPP equation with 2.0- and 1.5-L fill volumes were 0.534 and 0.624, respectively.

Table 2:

Regression coefficients between aIPP and variables by multiple linear regression analysis and newly derived equation for eIPP with 2.0-L fill volume and 1.5-L fill volume.

	2.0-L fill volume				1.5-L fill volume
Variables (n = 100)	Coefficients	t	P-value	R 2	Coefficients	t	P-value	R 2
Waist	0.193	6.784	<.001	0.534	0.218	9.092	<.001
Distance from decubitus plane to mid-axillary line	−0.907	−8.786	<.001		−0.850	−9.801	<.001	0.624
Constant	8.613	3.318	.001		4.127	1.893	.061
Equation	eIPP (cmH2O) = Waist (cm) * 0.193 – distance from decubitus plane to mid-axillary line * 0.907 + 8.613				eIPP (cmH2O) = waist (cm) * 0.218 − distance from decubitus plane to mid-axillary line * 0.850 + 4.127

Measurements taken for each volume in the development group (n = 100). Statistical test: stepwise linear regression analysis.

Validation of new equations and comparisons with existing equations

The mean aIPP measured using the Durand method in the validation dataset was 16.5 ± 4.3 cmH₂O with a 2.0-L fill volume and 14.6 ± 4.2 cmH₂O with a 1.5-L fill volume. The bias between eIPP and aIPP was small, at only −0.24 cmH₂O (95% CI −0.97–0.43) for 2.0 L (Fig. 3A) and −0.10 cmH₂O (95% CI −0.78–0.51) for 1.5 L. The SD of the absolute values of differences in the eIPP equation was 2.59 cmH₂O (95% CI 1.87–3.20) for the 2.0-L fill volume (Table 3) and 2.50 cmH₂O (95% CI 1.62–3.26) for the 1.5-L fill volume, indicating good precision. The value of 1 − P20 for the eIPP was 26% (Table 3) and 27% for 2.0 L and 1.5 L, respectively, representing fair accuracy.

Figure 3:

Difference between IPP measured by the Durand method (aIPP) and that measured by (A) the new equation (eIPP), (B) the Castellanos equation, (C) the Scanziani equation and (D) the De Jesus Ventura equation with a 2-L fill volume in the validation dataset of 100 patients. Statistical test: Bland–Altman analysis.

Table 3:

Performance of eIPP and other existing equations for estimating IPP with Durand as the reference method with a 2.0-L fill volume.

Equation	Validation dataset (n = 100)	P-value	Male (n = 68)	Female (n = 32)	P-value	Age <65 years (n = 81)	Age ≥65 years (n = 19)	P-value
Bias, median difference (95% CI)
eIPP	−0.24 (−0.97–0.43)	Ref.	−0.11 (−1.03–0.69)	−0.50 (−1.91–1.27)	.686	0.01 (−0.62–0.56)	−1.31 (−4.18–1.55)	.275
Castellanos equation [23]	−2.48(−3.70 to −1.31)	.001	−1.10(−2.62–0.28)	−5.42 (−7.91 to −2.31)	.039	−2.55(−4.24–0.92)	−2.19(−5.80–0.43)	.745
Scanziani equation [19]	−2.25 (−3.00 to −1.43)	.003	−2.07 (−3.36–1.08)	−2.63 (−3.85 to −0.95)	.588	−1.98 (−2.69 to −1.07)	−3.38 (−6.02 to −0.94)	.216
De Jesus Ventura equation [21]	0.98 (0.09–1.88)	.068	1.88 (0.48–2.92)	−0.94 (−2.27–0.71)	.020	1.43 (0.65–2.34)	−0.98 (−3.86–1.29)	.039
Precision, SD of the difference (95% CI)
eIPP	2.59 (1.87–3.20)	Ref.	2.54 (1.40–3.24)	2.72 (1.35–3.73)	.549	2.10 (1.40–2.75)	3.91 (1.77–5.27)	.059
Castellanos equation [23]	4.36 (3.45–5.18)	<.001	2.98 (2.36–3.43)	5.98 (4.92–7.69)	.059	4.51 (3.31–5.74)	3.71 (2.29–4.12)	.863
Scanziani equation [19]	2.79 (2.25–3.26)	.017	2.83 (2.11–3.54)	2.74 (1.55–3.52)	.725	2.39 (1.98–2.79)	3.96 (2.23–5.27)	.157
De Jesus Ventura equation [21]	2.86 (2.42–3.23)	.015	3.06 (2.25–3.34)	2.38 (1.81–2.71)	.373	2.77 (2.12–3.18)	3.13 (1.63–4.02)	.196
Accuracy, % (95% CI), 1 − P20a (%)
eIPP	26.0 (17.5–34.7)	Ref.	23.5 (14.7–38.6)	31.3 (11.7–50.4)	.412	23.5 (14.2–31.2)	36.8 (12.0–58.0)	.231
Castellanos equation [23]	56.0 (46.7–65.5)	<.001	51.5 (40.1–60.6)	65.6 (48.2–82.7)	.183	59.3 (48.8–72.7)	42.1 (10.7–69.2)	.175
Scanziani equation [19]	49.0 (38.5–59.0)	<.001	45.6 (32.8–62.9)	56.3 (35.7–77.1)	.320	45.7 (35.3–56.3)	63.2 (35.1–84.4)	.170
De Jesus Ventura equation [21]	40.0 (30.9–50.0)	.011	41.2 (26.9–51.3)	37.5 (17.3–55.3)	.726	38.3 (26.3–47.5)	47.7 (29.9–64.3)	.466

Statistical tests: block analysis of variance was used to compare bias and precision with a 2.0-L fill volume in the total validation dataset with the data presented as the difference between the value of existing equations and eIPP (P); McNemar's test was used to compare accuracy between the value of existing equations and eIPP (P).

^a1 – P20 accuracy was calculated as the percentage of estimates that differed from the measured IPP by >20%.

When the fill volume was 2.0 L, the mean aIPP measured using the Durand method in the validation dataset was 16.5 ± 4.3 cmH₂O. The eIPP values from the newly derived equation slightly underestimated IPP (P = .649). In contrast, the de Jesus Ventura equation slightly overestimated IPP (P = .061). IPP values estimated using the Castellanos equation and the Scanziani equation significantly underestimated IPP (P < .001 for both).

Concerning the performance of equations with a 2.0-L fill volume, the bias of the new equation for eIPP (−0.24 cmH₂O) was smaller than for the Castellanos equation (−2.48 cmH₂O, P = .001), the Scanziani equation (−2.25 cmH₂O, P = .003) and the de Jesus Ventura equation (0.98 cmH₂O, P = .068) using aIPP as the reference (Table 3, Figs. 2 and 3B–D). In terms of bias in subgroups, the bias of the Castellanos equation was lower in the male subgroup and the bias of the De Jesus Ventura equation was higher in the male and <65-years subgroups. No differences in terms of bias in subgroups with male or female and <65 or ≥65 years were found in the new equation and the Scanziani equation. The SD for the absolute values of differences by the new equation (2.59 cmH₂O) was smaller than that found by the Castellanos equation (4.36 cmH₂O, P < .001), the Scanziani equation (2.79 cmH₂O, P = .017) and the de Jesus Ventura equation (2.86 cmH₂O, P = .015) (Table 3, Fig. 2). In terms of percentage accuracy, the 1 − P20 of the new equation for eIPP (26.0%) was lower than with the Castellanos equation (56.0%, P < .001), Scanziani equation (49.0%, P < .001) and de Jesus Ventura equation (40.0%, P = .011) (Table 3). No differences in terms of precision and accuracy in subgroups with male or female and <65 or ≥65 years were found, no matter what equations were used for estimating IPP.

Figure 2:

Bias and precision of the new equation and other existing equations with a 2-L fill volume (n = 100). Statistical test: box plot. The box represents the median and IQR of data and the whiskers represent the maximum and minimum values. *Denotes extremes in the box plot that were >3 box lengths away from the upper quartile or the lower quartile.

DISCUSSION

We developed two novel equations to predict IPP with 2.0-L and 1.5-L fill volumes in adult PD patients. Our results showed that eIPP values using waist circumference and distance from the decubitus plane to the mid-axillary line were close to the aIPP value measured by the Durand method. The bias between eIPP and aIPP was very small (i.e. −0.24 cmH₂O and −0.10 cmH₂O for 2.0 L and 1.5 L, respectively). Compared with the performance of existing equations for a 2.0-L infused volume [19, 21, 23], the newly derived eIPP equation was more reliable, considering its lower bias, improved precision and higher accuracy.

The mean IPP reported by previous studies was 13–17 cmH₂O in adults [6, 22, 23, 33, 34]. An IPP >18–20 cmH₂O is usually considered pathological [23, 35] since it often becomes clinically symptomatic and corresponds with a decrease of >20% in pulmonary vital capacity [36]. In our study, the mean IPP was 14.6 cmH₂O at 1.5 L of dialysate and 16.5 cmH₂O at 2.0 L of dialysate, and the prevalence of elevated IPP at 2.0 L (if defined as >18 cmH₂O) was 31.5%, which is comparable to previous results [6, 18, 22, 23, 26]. The IPP increased 1.9 cmH₂O while the IPV increased from 1.5 L to 2.0 L in our cohort, which is comparable to 1.33–2.1 cmH₂O [15, 18, 19, 22], as shown by previous data. This allows us to individually tailor dialysis prescriptions for new patients. As shown by our previous work, there is a relatively low prevalence of abdominal wall complications, even in the setting of urgent-start PD [9].

Whole-body anthropometry measurements such as BMI and BSA are generally considered as major contributors to IPP [18–21, 23]. In contrast, the relationship between waist circumference and IPP was only observed in two previous studies [23, 26]. From our data, we observed positive associations between both whole-body and abdominal anthropometry data and IPP by univariate Spearman's correlation analyses. Furthermore, we found that only anthropometry measurements related to the size of the peritoneal cavity were independently associated with the IPP, including waist circumference and distance from the mid-axillary line to the decubitus plane. There are several causes for our findings. First, patient's perception of pressure derived from the traction of local abdominal wall tissue is directly related to IPP. This is why the amount of dialysate is always tailored to fit the patient's subjective feelings such as abdominal pain or bloating in traditional clinical practice [37]. A larger waist might be a consequence of lower abdominal musculature tone and greater abdominal wall depth, reasonably predictive of a higher IPP. Second, waist circumference and waist:hip ratio are representative of visceral fat mass as surrogate measures of abdominal obesity [38–40], which are also stronger predictors of all-cause and cardiovascular deaths than BMI [40–43]. In the setting of PD, an IPV of 2.0 L per se would reduce vital capacity between 9% and 20% [44]. More visceral fat mass may make the space for infused fluid smaller and thus predict a higher IPP. In contrast, the distance from the mid-axillary line to the decubitus plane as a measure of skeleton frame is negatively correlated with the IPP. A greater distance from the mid-axillary line to the decubitus plane may reflect a larger peritoneal cavity, thus reducing the IPP.

Apart from abdominal anthropometric data, components of metabolic syndrome such as diabetes and serum triglycerides and inflammation marker (hs-CRP) were also predictive of the IPP by univariate analyses. However, their associations with IPP disappeared after adjustment for abdominal anthropometric data, which can be explained by confounding effects, as these variables are closely associated with visceral fat mass as reported by previous studies [45–47]. A relationship between IPP and gender was not observed in our study. This matter is controversial, as several studies have reported inconsistent associations between IPP and gender [21–23].

This study has several strengths. To start, this is the first study to consider local anthropometric data relating to the volume of the peritoneal cavity when estimating the IPP. Compared with several existing equations for a 2.0-L infused volume, our newly derived equations have good performances in that they have smaller biases, better precision and greater accuracy. These data verified the validity of abdominal anthropometry data. Our sample size is relatively large compared with previous studies, which usually included <100 subjects. Patients were randomly distributed in the development and validation datasets according to BMI. This precludes a choosing bias in the validation of the new equation.

We are also aware of some limitations. First, similar to previous research, our equation was developed to estimate the IPP in the supine position rather than in a sitting and upright position. The current equations cannot predict the varying IPP in daily activities as reported [48]. The IPP also changed along with the dialysis duration [23, 49]. We need to design novel techniques to reflect the real-time IPP in a convenient and simple way. Until then, estimating the IPP with any specific equations can be done in the static state. Second, although abdominal anthropometric data were measured, we cannot calculate the volume of the abdominal cavity based on these variables. Sigogne et al. [18] measured the liver and kidney volume with Osiri X software and a contouring technique on serial sections of computed tomography scans. If the volume of most solid organs in the abdomen can be calculated by image techniques, organ-free volumes for containing the solution may be estimated. Third, only clinically stable patients with no history of hernia, leakage or hydrothorax were included in our study. Patients with these complications resumed PD after surgical repair. They need to be monitored more carefully since there is a higher risk for repeat complications [5, 9]. Fourth, the measurement of IPP is different in patients with polycystic kidney or polycystic liver, because of the enlargement of organs. It is suggested to use the equation established in polycystic kidney patients. Finally, although the Asian population is ethnically different from other populations and the clinical characteristics such as BMI in this study were compared with previous studies with other ethnicities [18, 19, 50, 51], we still should validate our new equations in a larger, multicentre sample population.

In summary, we developed two new equations, mainly from abdominal anthropometric measurements, for estimating the IPP at infused volumes of 1.5 and 2.0 L. Our equations were verified to be a feasible and reliable approach to estimate IPP and therefore can assist in providing individualized prescriptions for the PD population. The generalizability of the new equations should be investigated in a larger PD population and with different IPVs. Whether or not the eIPP based on our equations is valuable for monitoring the occurrence of abdominal wall complications needs to be explored in the future.

FUNDING

This work is supported by Scientific Research Project of Capital Health Development (2020-2-4079), New Century Excellent Talents from Education Department of China and CAMS Innovation Fund for Medical Sciences (2019-I2M-5-046) and a Clinic Research Award from the ISN GO R&P Committee for analysis and interpretation of the data.

AUTHORS’ CONTRIBUTIONS

J.D., T.M. and X.L. were responsible for the research idea and study design. T.M., X.L., J.H., D.S., H.W., T.L., Y.Z., N.A. and X.X. were responsible for data acquisition. X.L., T.M., X.X. and J.D. were responsible for statistical analysis. X.L., T.M. and J.D. were responsible for manuscript drafting or revision. J.D. was responsible for supervision or mentorship. Each author contributed important intellectual content during manuscript drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved.

DATA AVAILABILITY STATEMENT

Data described in the manuscript, code book and analytic code will not be made available because of laws governing the management of China's human genetic resources.

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no conflicts of interest. The results presented in this article have not been published previously in whole or part.