External Validation of Equations to Estimate Resting Energy Expenditure in Critically Ill Children and Adolescents with and without Malnutrition: A Cross-Sectional Study

We evaluated the validity of sixteen predictive energy expenditure equations for resting energy expenditure estimation (eREE) against measured resting energy expenditure using indirect calorimetry (REEIC) in 153 critically ill children. Predictive equations were included based on weight, height, sex, and age. The agreement between eREE and REEIC was analyzed using the Bland–Altman method. Precision was defined by the 95% limits of the agreement; differences > ±10% from REEIC were considered clinically unacceptable. The reliability was assessed by the intraclass correlation coefficient (Cronbach’s alpha). The influence of anthropometric, nutritional, and clinical variables on REEIC was also assessed. Thirty (19.6%) of the 153 enrolled patients were malnourished (19.6%), and fifty-four were overweight (10.5%) or obese (24.8%). All patients received sedation and analgesia. Mortality was 3.9%. The calculated eREE either underestimated (median 606, IQR 512; 784 kcal/day) or overestimated (1126.6, 929; 1340 kcal/day) REEIC compared with indirect calorimetry (928.3, 651; 1239 kcal/day). These differences resulted in significant biases of −342 to 592 kcal (95% limits of agreement (precision)−1107 to 1380 kcal/day) and high coefficients of variation (up to 1242%). Although predicted equations exhibited moderate reliability, the clinically acceptable ±10% accuracy rate ranged from only 6.5% to a maximum of 24.2%, with the inaccuracy varying from −31% to +71.5% of the measured patient’s energy needs. REEIC (p = 0.017) and eREE (p < 0.001) were higher in the underweight compared to overweight and obese patients. Apart from a younger age, malnutrition, clinical characteristics, temperature, vasoactive drugs, neuromuscular blockade, and energy intake did not affect REEIC and thereby predictive equations’ accuracy. Commonly used predictive equations for calculating energy needs are inaccurate for individual patients, either underestimating or overestimating REEIC compared with indirect calorimetry. Altogether these findings underscore the urgency for measuring REEIC in clinical situations where accurate knowledge of energy needs is vital.


Introduction
Accurate determination of resting energy expenditure (REE) in critically ill patients is vital because underfeeding and overfeeding are both associated with undesirable consequences. Cross-sectional and longitudinal studies have shown that mechanically ventilated children do not increase their metabolic rate during the acute phase of critical illness [1].

Clinical Data
At the time of each metabolic measurement, the admission diagnosis, temperature, blood pressure, heart rate, sedation level by Ramsey scale, and main sedatives and vasoactive agents or inotropes were recorded. The last recorded temperature on a patient's vital signs flowchart just before the REE IC measurement was documented. The severity of illness was assessed using the PRISM-III and the PELOD-2 scores [31], and the amount of care was assessed using the Therapeutic Intervention Scoring System (TISS) [32]. The ventilatory settings at the time of the measurement and the route of nutrition support, and the total calories received for the 24-h period before metabolic measurement were also recorded. Energy intake was calculated from recorded intakes of enteral or parenteral nutrition and glucose-containing maintenance fluids. Underfeeding and overfeeding were defined according to the European Society for Clinical Nutrition and Metabolism (ESPEN) guidelines as intakes of <70% or >110% of REE IC , respectively [33].

Anthropometry
The following anthropometric parameters were identified: age, sex, actual weight, ideal weight, height, and body mass index (BMI). Weight was measured using calibrated electronic bed scales. Ideal weight was defined as the weight for the 50th percentile of the actual height of each patient. BMI was calculated as kg/m 2 . Standard deviations scores, known as z-scores, of weight, height, and BMI for sex and age were calculated using WHO and CDC calculators [34]. Malnutrition indices were derived from the BMI for age and sex z scores obtained at admission. Underweight was defined as BMI for sex and age z-score < −1.644, normal weight as −1.644 ≤ BMI z-score < 1.036, overweight as 1.036 ≤ BMI z-score < 1.644, and obesity as BMI z-score ≥ 1.644.

Indirect Calorimetry
An integrated gas exchange module (E-COVX) into the ventilator (Carescape R860; GE Healthcare, Milwaukee, WI, USA) was used to measure REE IC through indirect calorimetry on PICU day 3 or 4. This module is able to reliably record spirometry and metabolic indices as early as 5 min after suctioning at different modes of ventilation [26,27]. It has no mixing chamber and sampling takes place with every breath. It has a fast differential paramagnetic O 2 and infrared CO 2 analyzer and a pneumotachograph housed in a connector, which measures inspired and expired volumes. In the P-Lite (15-300 mL) or D-Lite (>300 mL) flow sensor, located proximate to the Y-piece to the patients' ET tube the flow measurement is based on the pressure drop across a special proprietary turbulent flow restrictor. It uses mathematical integration of flow and time-synchronized continuous gas sampling to provide data. The gas sample is continuously drawn from the connector to the gas analyzer unit of the module. Both O 2 and CO 2 measures are based on the side-stream principle. The E-COVX relies on tidal volume measurement for VO 2 calculation. The pneumotachograph derives the tidal volume from the pressure difference across a fixed orifice, potentially influenced therefore by acute changes of resistance in the spirometry tubing and undetected leaks in the system. We consistently used a heat-and moistureexchange filter alone, avoiding heated water bath humidification, followed by regular checks on the spirometry tubing and checks for tidal volume consistency between the module and the ventilator.
Measurements were made between 9 am and 12 pm when there had been a minimum of 45 min with no major physical activity, such as physiotherapy or dressing change. After an initial 10-min stabilization period, REE IC was measured for 30 min, during which time there was no interference with the child. The module uses the modified Weir formula (REE IC (kcal/day) = [3.941 × VO 2 + 1.106 × VCO 2 ] × 1440 and displays a 5-min average for REE IC but can display the 1-min averages with the S/5 Collect 1.0 software (Datex-Ohmeda, GE Healthcare, USA). Steady state was defined as a period of at least 5 min with less than 10% fluctuation in VO 2 and VCO 2 , and less than 5% fluctuation in respiratory quotient (RQ), which is the ratio of VCO 2 : VO 2 . Measurements with RQ outside the physiologic range (>1.3 or <0.67) were excluded.
Basal metabolism was calculated based on the Schofield equation. The metabolic state for each patient was determined using the ratio of measured REE IC to eREE based on the Schofield equation, as has been previously suggested [35][36][37]. Patients were classified in the following metabolic patterns: normometabolic when REE IC /eREE Schofield = 90-110%, hypometabolic when REE IC /eREE Schofield < 90%, and hypermetabolic when REE IC /eREE Schofiled > 110%.

Statistical Analysis
The normality of the distribution was assessed using the Shapiro-Wilk test. Descriptive data are reported as means and standard deviation (SD) or median and interquartile range (IQR) in case of skewed distributions, or as frequencies and percentages when appropriate. The accuracy of the eREE compared to REE IC measured by indirect calorimetry was assessed through the calculation of bias and precision. Bias was defined as the mean difference between the measurements obtained from the eREE and REE IC . Precision was defined by the 95% limits of the agreement including both systematic (bias) and random error. The percentage of predicted values of an equation within 10% of REE IC was considered a measure of accuracy on a cohort or sub-cohort level. The relative variability (dispersion) and repeatability were assessed by calculating the coefficient of variation (CV) which is the ratio of the standard deviation to the mean of the population. The reliability was assessed by the intraclass correlation coefficient (ICC), calculated using the two-way mixed (Cronbach's alpha). ICC was interpreted as follows: below 0.50: poor; between 0.50 and 0.75: moderate; between 0.75 and 0.90: good; above 0.90: excellent [38]. A linear regression model (backward method) was adopted to examine whether any of the recorded anthropometric, clinical, and nutritional variables are independently associated with REE IC . We first used univariate models to test if any of the studied variables were related to REE IC , with just one explanatory variable at a time; afterward, all variables that had shown a relaxed p-value of less than or equal to 0.1 were included in the multivariable models. A two-sided significance level of 0.05 was used for statistical inference. Statistical analysis software (version 28; SPSS, Chicago, IL, USA) was used for all analyses and GraphPad Prism 9.0 (GraphPad Software, Inc., San Diego, CA, USA) was used for the Bland-Altman analyses and illustrations.

Study Population
During the study period, 735 patients were admitted to the PICU, of which 176 were eligible for inclusion. However, 23 patients were not enrolled due to logistical reasons (n = 12), technical reasons (n = 7), or no informed consent (n = 4). Demographic, anthropometric, clinical, and metabolic characteristics are shown in Table 1. Less than half of the patients (n = 69; 45.1%) had a BMI within the normal range for their sex and age. Thirty (19.6%) patients were underweight (19.6%), and fifty-four were overweight (10.5%) or obese (24.8%). All patients received sedation and analgesia. Mortality was 3.9%. Nutritional support was provided enterally (90.9%) or parenterally (9.1%). Patients' feeding status on PICU day three revealed that two-thirds of the patients were either underfed (39.8%) or overfed (27.6%). Normal weight patients received targeted nutrition in only 36.7%, while 36.7% were underfed and 26.7% were overfed. In contrast, 55.6% of obese patients were underfed whereas 36% of underweight patients were overfed (Table 1).

Performance of Predictive Equations
The calculated eREE either underestimated (median 606, IQR 512; 784 kcal/day) or overestimated (1126.6, 929; 1340 kcal/day) REE IC compared with indirect calorimetry  Table 2. The calculated age and sex-specific RDA, grossly overestimated REE IC )median 1320 (IQR 880; 2365) kcal/day). These differences resulted in significant biases of −342 to 592 kcal (95% limits of agreement (precision) −1107 to 1380 kcal/day). Even predictive equations with small bias (Molnar, Caldwell-Kennedy, Henry (Oxford), Meyer) exhibited extended dispersion of values as visualized by the 95% limits of agreement in the Bland-Altman plots ( Figure 1). Compared to indirect calorimetry, old or new equations, irrelevant to the established age, nutrition, race, or illness-related status, presented a large bias and small precision, indicated by the wide 95% limits of agreement in the Bland-Altman plots ( Figures S1-S3).
Paired eREE-REE IC differences were significant for most predictive equations (Wilcoxon matched-pairs signed rank test, medians of differences −282 to +627, p < 0.002) except for the Molnar, Caldwell-Kennedy, Henry (Oxford), Meyer equations (−8 to +137, p > 0.05). These equations, however, were also inaccurate, presenting a wide dispersion of values as expressed by a high coefficient of variation (809-1242%), in accordance with their high bias and limits of agreement ( Table 2).
The equations' reliability, as assessed by the ICC, although significant (p < 0.001), varied at moderate levels between 0.51 and 0.74 and was consistent across sub cohorts of obese, overweight, and underweight patients (Cronbach's alpha, Table 3).
The calculated age and sex-specific RDA, grossly overestimated REEIC )median 1320 (IQR 880; 2365) kcal/day). These differences resulted in significant biases of −342 to 592 kcal (95% limits of agreement (precision) −1107 to 1380 kcal/day). Even predictive equations with small bias (Molnar, Caldwell-Kennedy, Henry (Oxford), Meyer) exhibited extended dispersion of values as visualized by the 95% limits of agreement in the Bland-Altman plots (Figure 1). Compared to indirect calorimetry, old or new equations, irrelevant to the established age, nutrition, race, or illness-related status, presented a large bias and small precision, indicated by the wide 95% limits of agreement in the Bland-Altman plots ( Figures S1-S3). Despite the moderate reliability, the 10% accuracy rate ranged from 6.5% to a maximum of 24.2%, and it was significantly lower than an expected minimum accuracy overall and across nutrition status sub-cohorts (Table 3). Inaccuracy profile varied from underestimation (White, median −31%, IQR −44%; −9.5%) to overestimation (Lazzer, median 71.5%, IQR 28.6%; 138%) of the patient's energy needs (p < 0.001) (Figure 2). and it was significantly lower than an expected minimum accuracy of 50%.

Malnutrition and Factors Independently Associated with REEIC
Measured by indirect calorimetry, REEIC (kcal/kg/day) was higher in the underweight and lower in the obese compared to other sub-cohorts (p = 0.017). All predicted equations also calculated higher kcal/kg/day in the underweight compared to overweight and obese patients (p < 0.001) (Figure 3).

Malnutrition and Factors Independently Associated with REE IC
Measured by indirect calorimetry, REE IC (kcal/kg/day) was higher in the underweight and lower in the obese compared to other sub-cohorts (p = 0.017). All predicted equations also calculated higher kcal/kg/day in the underweight compared to overweight and obese patients (p < 0.001) (Figure 3).    In a linear regression model (stepwise, backward method), only a younger age (Beta −0.49, p < 0.001) was independently associated with the measured REEIC. None of the BMI nutrition status (overweight, obesity), the severity of illness (PRISM, TISS, PELOD), diagnostic category, outcome, temperature, heart rate, lactate, vasoactive drugs, neuromuscular blockade, or energy intake were independently associated with the REEIC.

Discussion
The accurate determination of energy needs in critically ill children is vital because underfeeding and overfeeding are both associated with undesirable consequences. Alt- In a linear regression model (stepwise, backward method), only a younger age (Beta −0.49, p < 0.001) was independently associated with the measured REE IC . None of the BMI nutrition status (overweight, obesity), the severity of illness (PRISM, TISS, PELOD), diagnostic category, outcome, temperature, heart rate, lactate, vasoactive drugs, neuromuscular blockade, or energy intake were independently associated with the REE IC .

Discussion
The accurate determination of energy needs in critically ill children is vital because underfeeding and overfeeding are both associated with undesirable consequences. Although IC is considered the gold standard for assessing REE in ICU patients, several predictive equations, developed from measured energy expenditure based on various numbers of healthy non-hospitalized subjects, are commonly used in clinical practice. In this study, we evaluated commonly used previously validated equations and found that even the most accurate equations had an unacceptably high error. We showed that recommended or not, PICU-related or not, the older or newly predictive equations presented large biases and small precisions, as indicated by the wide 95% limits of agreement in the Bland-Altman plots, significant paired differences, and high coefficients of variation. We also showed that although sixteen predicted equations exhibited moderate reliability, the clinically acceptable 10% accuracy rate ranged from only 6.5% to a maximum of 24.2%, with the inaccuracy varying from −31% to +71.5% of the measured patients' energy needs. Finally, we demonstrated that, apart from a younger age, malnutrition, clinical characteristics, temperature, vasoactive drugs, neuromuscular blockade, and energy intake did not affect REE IC and thereby eREE.
A novel finding of this study is that the inaccuracy of the assessed predictive equations did not correlate with the established time (old or new), age range (pediatric, adult), malnutrition status, race, illness-related status (healthy, PICU), or recommendation by scientific societies (Schofield, WHO, IOM). For predicting energy requirements, the Schofield [5] and FAO/WHO/UNU [3] equations have been previously recommended for the healthy pediatric population [39], while in a population with obesity, the Molnár [11] and Dietz [9] equations performed most accurately. For patients receiving mechanical ventilator support, the Harris-Benedict predicted more accurately than other equations, but with a wide error range (±500 kcal) [17]. In our critically ill, mechanically ventilated patients, predicted equations either underestimated or overestimated REE, compared with measured REE IC . All predictions presented significant matched paired eREE-REE IC differences, a wide dispersion of values as expressed by high coefficients of variation, significant biases of −342 to 592 kcal, and poor precision (−1107 to 1380 kcal/day). Most of the equations overestimated REE IC , erroneously calculating higher energy needs of critically ill patients. Findings of previous studies using indirect calorimetry support our conclusion that children do not become hypermetabolic during critical illness [36] and that improved PICU-specific prediction methods are still imprecise in critically ill children [23,[40][41][42][43].
Our data suggest that simple predictive equations may lead to overfeeding in critically ill children and less often to underfeeding. A U-shaped association between mortality and energy intake revealed the importance of personalized energy support and the need to prevent overfeeding and underfeeding [44]. Two recent meta-analyses showed a reduction in ICU mortality when feeding protocols were based on REE IC [45] compared to eREE [46]. Nutrition guidelines recommend measuring REE using a validated indirect calorimeter to guide nutritional support in critically ill infants and children after the acute phase [47]. Alternatively, the Schofield equation is recommended to estimate REE [47], which we showed to be one of the most inaccurate. Imprecise predictive equations that overestimated REE IC more than others were the RDA (95% limits of agreement −1101 to 2585 kcal/day), Lazzer (−196 to 1380 kcal/day), IOM (−593; 1011 kcal/day), Kaneko (−549 to 967 kcal/day), Schofield H-W (−652 to 1021 kcal/day), and Dietz-(598 to 959 kcal/day) equations. Although the FAO/WHO/UNU, Harris-Benedict, Maffeis, Lawrence, and Muller equations' overestimation bias was smaller, they were inaccurate with wide 95% limits of agreement.
Finally, the two equations that mostly underestimated REE IC were the White (−1107 to 422 kcal/day) and Mifflin (−966.8 to 575 kcal/day) equations.
In accordance with the results of the Vazquez Martinez study in the early postinjury period [17], we found the Caldwell-Kennedy equation to be among the four less inaccurate predictors of energy expenditure in ventilated, critically ill children. However, even the four predictive equations with the smallest bias, Molnar (−32 kcal/day), Caldwell-Kennedy (44 kcal/day), Henry (Oxford) (−47 kcal/day), and Meyer (47 kcal/day), exhibited extended dispersion of values as visualized by a high coefficient of variation (809-1242) and wide limits of agreement (+539.55; 1378.03 kcal/day). In the absence of IC, American Society for Parenteral and Enteral Nutrition (ASPEN) guidelines suggested that a published predictive equation or a simplistic weight-based equation (25-30 kcal/kg/d) be used in adults to determine energy requirements [48]. However, if predictive equations are used to estimate the energy need, hypocaloric nutrition (below 70% of eREE) should be preferred over isocaloric nutrition for the first week of ICU stay as per ESPEN guidelines [33].
In our series, more than half of the patients were malnourished, whereas two-thirds were underfed or overfed. In addition, both indirect calorimetry and predicted equations calculated higher kcal/kg in the underweight compared to overweight and obese patients. Following the same trend, obese patients were underfed (70.4%), whereas 36% of underweight patients were overfed. It has been suggested that patients who are at high nutrition risk or severely malnourished should be advanced to provide >80% of REE IC or eREE and protein within 48-72 h to achieve the clinical benefit of early enteral nutrition while monitoring for refeeding syndrome [48]. Hypocaloric parenteral nutrition dosing (80% of eREE) with adequate protein (≥1.2 g protein/kg/d) should also be considered in high-risk or severely malnourished patients requiring parenteral nutrition over the first week in ICU [48]. Regarding obesity, the guidelines suggest that the goal of enteral nutrition should not exceed 65%-70% of the target REE IC [48]. Personalized nutritional adjustments may impact PICU length of stay, readmission rates, quality of life [49], and long-term rehabilitation success [50]. Scientific societies recommend measuring REE by IC in malnourished children and/or suspected altered metabolism. According to these criteria, more than 70% of PICU patients are candidates for IC measurement [51]. Our finding that <25% of the equations predicted REE IC within ±10% of the indirect calorimetry REE IC exaggerates the results of a systematic review study, showing that no equation predicted REE IC within ±10% in >50% of observations [52].
Most of our patients were hypometabolic, in accordance with previously published data (5,6,(15)(16)(17). Several factors have been implicated to explain the hypometabolism of critically ill children, such as coma, mechanical ventilation, analgesia, sedation, neuromuscular blockade, and malnutrition. It is the first time, however, to demonstrate that none of the malnutrition status, the severity of illness, diagnostic category, outcome, temperature, heart rate, lactate, vasoactive drugs, neuromuscular blockade, or energy intake were independently associated with the REE IC inaccuracy. In agreement with findings of an adult study in critically ill medical patients [53], we showed that only a younger age is independently associated with indirect calorimetry measurements in mechanically ventilated children. Accordingly, except for age, none of the estimated nutritional or clinical confounders might indirectly affect the REE IC -eREE difference. This hypothesis is further supported by the fact that PICU-related equations did not perform better than other predictive equations.
One of the limitations of this study is the small sample size, although it is in the upper range of similar studies, including sixteen predictive equations, older, recent, adult, pediatric, PICU-related, and nutrition status-related equations. In addition, the timing of the IC measurements in this prospective cross-sectional study only reflects the acute and not the recovery metabolic phase of illness. According to the ESPEN guidelines, every critically ill patient staying for more than 48 h in the ICU should be considered at risk for malnutrition [33]. We measured REE on ICU Day 3 or 4 since it has been previously shown that non-inhibitable endogenous energy is produced in the acute phase of critical illness due to a catabolic state [50]. Since the non-measurable, adapted to acute illness endogenous effect dissipates by Day 4 [54], it is recommended to commence early enteral nutrition within 24 h of admission [55], and to increase it in a stepwise fashion until the goal for delivery is achieved using a feeding protocol [47], to avoid overfeeding and mitochondrial exhaustion by targeting REE IC during the acute stress period [49,56]. Adult guidelines also recommend that hypocaloric nutrition (not exceeding 70% of REE IC ) should be administered in the early phase of acute illness and that isocaloric nutrition should be progressively implemented after the early phase of acute illness [33]. Because of the unpredictable effects of a critical illness on metabolism, the considerable variation in REE, and the progressive hypermetabolism, IC should be used daily in assessing nutrition in ICU patients [57]. After the acute phase, energy intake should account for energy deficits, physical activity, or exercise, and growth [47]. Recently developed self-calibrating and simple-to-operate instruments, with implemented artificial intelligence, have built-in algorithms for the detection and deletion of aberrant periods of measurements resulting from breathing variability [58]. Future developments of metabolic cart technology to reliably monitor REE IC continuously in states of respiratory and circulatory instability, using various ventilatory settings, including non-invasive ventilation, are expected to facilitate the daily application of IC in an intensive care setting.

Conclusions
All available prediction equations for calculating energy needs are inaccurate for individual patients, either underestimating or overestimating REE compared with indirect calorimetry. Apart from a younger age, malnutrition, clinical characteristics, temperature, vasoactive drugs, neuromuscular blockade, and energy intake did not affect REE IC and thereby the accuracy of the predictive equations. Sixteen predictive equations may result in under-or overfeeding and cannot substitute for indirect calorimetry measurement of energy expenditure in guiding the personalization of nutrition delivery in pediatric intensive care patients.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/nu14194149/s1, Table S1: Predicted energy expenditure equations compared to indirect calorimetry for calculating energy expenditure in critically ill children; Figure S1: Bland-Altman plot whereby estimated by predicted equations' resting energy expenditure (eREE) is compared to REE measured by IC (REE IC ) at ICU Day-3 or 4. A. Harris-Benedict eREE compared to REE IC . B. Schofield (height and weight, WHO) eREE compared to REE IC . C. Mifflin eREE compared to REE IC . D. Muller eREE compared to REE IC . The solid line indicates the percentage of agreement bias (%) and the light shade with the fine dotted lines indicates the limits of agreement (bias ± (1.96 × SD) = precision). Dark shade represents the 95% confidence intervals of the mean (bias).; Figure S2: Bland-Altman plot whereby estimated by predicted equations' resting energy expenditure (eREE) is compared to REE measured by IC (REE IC ) at ICU Day-3 or 4. A. Maffeis eREE compared to REE IC . B. White (Equation (2)) eREE compared to REE IC . C. Institute for Medicine of the National Academies and Food and Nutrition Board (IOM) eREE compared to REE IC . D. Dietz eREE compared to REE IC . The solid line indicates the percentage of agreement bias (%) and the light shade with the fine dotted lines indicates the limits of agreement (bias ± (1.96 × SD) = precision). Dark shade represents the 95% confidence intervals of the mean (bias); Figure S3: Bland-Altman plot whereby estimated by predicted equations' resting energy expenditure (eREE) is compared to REE measured by IC (REE IC ) at ICU Day-3 or 4. A. FAO/WHO/UNU eREE compared to REE IC . B. Lazzer (Equation (1)) eREE compared to REE IC . C. Lawrence-3 eREE compared to REE IC . D. Kaneko eREE compared to REE IC . The solid line indicates the percentage of agreement bias (%) and the light shade with the fine dotted lines indicates the limits of agreement (bias ± (1.96 × SD) = precision). Dark shade represents the 95% confidence intervals of the mean (bias). References [2][3][4][5][6][7][8][9][10][11][12][13]15,[17][18][19][20]. Institutional Review Board Statement: The study was conducted in accordance with the principles of the Declaration of Helsinki (last revised guidelines from 2013), following the International Conference on Harmonization (ICH)/Good Clinical Practice (GCP) standards for studies involving humans [30] and the Ethics Committee of the Institutional Review Board approved the study (approval ID14494/2011/9-1-2012).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The datasets generated and analyzed during the current study are not publicly available because the database is very extensive and includes data from other studies complementary to this, but are available from the corresponding authors upon reasonable request.