We developed ultrasound-based normative values for kidney length in Central European children from birth to 19 years of age shown as age- and height-related percentiles, LMS-derived Z-scores, and quartile regression formulas. To our knowledge, our normative values are based on the largest group of children and adolescents published to date. Moreover, this is the first study on kidney length to include kidney function, which allowed us to exclude patients with impaired kidney function. This is also the first study on kidney length reporting LMS parameters for kidney size that can be used to calculate kidney length Z-scores and/or any given percentile of kidney length. For practical purposes, we developed percentiles (tables, curves) and simple formulas to calculate the 2.5th, 50th, and 97.5th percentiles of kidney size in relation to height.
Several previously published studies reported significant differences in kidney size between males and females. For example, in the longitudinal cohort study by Schmidt et al., boys had significantly larger kidney volumes than girls of all ages, and the sex difference was not due to body size . Differences in kidney length between boys and girls were also described by Scott et al. on a group of 560 healthy infants . However, the majority of studies reported no differences in kidney size between the sexes [1, 3, 7, 8, 15, 22]. We also did not observe significant differences in kidney length in relation to the height between girls and boys. However, boys over 15 years had significantly larger kidneys compared to girls of the same age (Fig. 3), which was most likely related to a pubertal growth spurt and greater height. This is similar to the differences in blood pressure values between the sexes. Furthermore, it is an additional argument for kidney size assessment based on height rather than age.
Predictors of kidney length
Kidney size correlates well with most of the currently used parameters of body size, including height, weight, BMI, and BSA. According to a study by Dinkel et al. on 325 children aged between 3 days and 15 years, the best predictor of kidney length was BSA . Similar results were obtained by Haugstvedt et al., who found a good correlation between kidney length and depth and variables like age, weight, height, and body surface area. However, BSA was the best predictor of kidney size .
Despite these significant correlations with BSA, it should be noted that the calculation of BSA is relatively cumbersome and requires measurements of both height and weight. In clinical practice, height and weight are measured directly in most patients. In contrast, BSA needs to be calculated, mostly for specific reasons only, e.g., as an index for GFR and left ventricular mass or drug dosing. In our study, there was also a significant correlation between BSA and kidney length (Fig. 4a); however, the correlation coefficient for BSA was still lower than the one for height.
While the relationship between age and kidney length is not linear (Fig. 1) and sex-dependent (adolescent boys have bigger kidneys than girls of the same age) (Table 2, Fig. 3), height was better correlated with kidney length with a higher r2 and no sex differences (Fig. 2).
Our findings are supported by the results of other studies in which height was the main predictor of kidney length. In the study published by Vujic et al., the strongest linear correlation coefficient was found between body length (height) and kidney length (r = 0.728 for the right kidney, r = 0.721 for the left kidney, and r = 0.724 for combined kidney length) and the combined kidney volume (r = 0.651) . Similar findings were recorded by Thapa et al. on a group of 272 pediatric subjects aged between 1 month and 15 years; the kidney length showed the strongest correlation with height and age . Height was also the main predictor of kidney length in children and adolescents in the study by Konus et al.; correlation coefficients with the kidney dimensions (longitudinal and transverse) were 0.94 and 0.86) . In the study by Coombs et al., height and weight were not measured, and kidney length was related to age only . Therefore, we conclude that from a statistical and clinical point of view, height seems to be the best predictor of kidney length irrespective of sex, age, and BMI. For clinical purposes, the median kidney length and its lower and upper limits (2.5th and 97.5th percentiles) can be predicted using simplified formulas (see above “Results”). For a more accurate assessment, the kidney length Z-scores/percentiles can be calculated using LMS parameters from Table 3.
The additional impact of BMI (besides age and height) on kidney length was further analyzed by multivariate analysis with BMI as a second independent predictor (in addition to age, height, or BSA). The relative impact of BMI on kidney length as an outcome measure was relatively small (up to 20%) compared to approximately 80% of variance accounted for by age, height, or BSA. The mixed linear model with age, sex, weight, height, and BMI showed that the most significant predictor of kidney length was height, followed by BMI and age (fixed effects), but the BMI had a negligible impact within age categories (random effect).
The potential differences between kidney size in the prone and supine positions may be clinically important. Michel et al. found that the maximum measured longitudinal kidney length was statistically significantly larger in the supine than the prone position (supine position, left: 8.0 cm; right: 7.7 cm; prone position, left: 7.9 cm; right: 7.6 cm; p < 0.001). Therefore, the authors recommended including prone kidney length measurements in addition to the supine measurements. However, this would complicate follow-up examinations, as the kidney length measurements can only be compared with the previous measurement in the same patient position .
Our study also found a statistically significant difference in kidney size (mean paired difference = − 0.64 mm, p < 0.001) between prone and supine positions. However, the absolute difference and effect size of this difference were minimal (Cohen’s D = 0.04), and the potential clinical significance may not be profound, especially given the existence of small intrinsic intra- and interobserver variability in sonographic length measurements [24, 25]. Therefore, we suggest measuring kidney length in any position in which kidney visualization is optimal.
The differences in kidney length between the right and left kidney are another debated topic in the literature. In the study published by Blane et al. on 34 infants, the left kidney was found to be longer than the right one by 3 mm; however, the standard error of this prediction was 4.4 mm . Scott et al. also confirmed that left kidneys were significantly longer and thinner than right ones. Although the differences in length and depth were highly significant, the confidence intervals showed that the scale of these differences was relatively small (about 1 mm) . Left kidneys were also statistically significantly longer by approximately 2 mm in the normograms published by Haugstvedt et al. based on a group of 46 children aged 0–16 years . Similar differences were observed in research by Kadioglu et al., including 292 children between 1 month and 18 years , and Michel et al., including 100 children from 6 months to 16 years . In contrast, some other studies did not find significant lateral differences in kidney size [15, 17, 26]. Our study found a significant statistical difference between right and left kidney length (mean paired difference = − 1.03 mm, p < 0.001). However, the absolute difference and its effect size were minimal (Cohen’s D = 0.06), not correlated with age or height. Therefore, we suggest that a statically significant lateral difference in kidney size is not clinically meaningful.
Comparison with other normative data
The comparison between our age-related normative data and other studies published by Rosenbaum et al. and Coombs et al. is shown in Fig. 5 [2, 17]. While we could not compare the curves statistically (in the absence of raw data in other studies), the visual analysis of curves reveals some potential differences in age-related kidney length between studies. However, the age-related normative data are difficult to compare as they may be influenced by the abovementioned additional factors such as sex and BMI. Moreover, our study has the largest population studied so far (n = 1,758) compared to Rosenbaum et al. (n = 203) and Coombs et al. (n = 940). To our knowledge, our study is the first to include LMS smoothing parameters for kidney length (in relation to height) in children, which makes the kidney length assessment more precise. Therefore, we believe that our study provides accurate and up-to-date normative data for kidney length in children.
One of the limitations of our study was the lack of kidney volume assessment. Although kidney volume theoretically correlates better with kidney weight and would be preferable to know in certain conditions such as autosomal dominant polycystic kidney disease, the kidney length is directly measurable using abdominal ultrasonography. In contrast, the estimation of kidney volume in two-dimensional sonographic examination requires more measurements, which increases variability, and involves a relatively complicated calculation based on a geometric assumption about the shape of the kidney, which may be time-consuming in clinical practice [25, 27,28,29]. Therefore, most published normative data for kidney size rely on kidney length rather than volume. Another limitation may be the lack of intra- and inter-observer variability assessment of kidney length assessment in our study. However, we believe that the number of patients and a modern statistical approach (estimation-based analysis, including effect size analysis, quantile regression, LMS smoothing) significantly limits the variability and excludes extreme values/outliers.