ABC | Volume 111, Nº1, July 2018

Original Article Oliveira et al Nonlinear dynamics in young adults with type 1 diabetes Arq Bras Cardiol. 2018; 111(1):94-101 In order to perform this stage, the volunteers were instructed to remain silent, awake, at rest, with spontaneous breathing for 30 minutes, in dorsal decubitus for HR measurement. All volunteers were instructed not to consume ANS stimulants, such as coffee, tea, soda, and chocolate, and not to perform physical activities 24 hours prior to the evaluation, in order not to interfere with cardiac autonomic modulation. The evaluations were performed individually in a roomwith temperature between 21°C and 23°C, and humidity between 40% and 60%, in the afternoon, between 1 pm and 6 pm to minimize circadian rhythm influences. 16 After HR measurement using a cardio-frequency meter, a thousand consecutive R-R intervals were selected from the period of greatest signal stability, and digital filtering was performed with a filter moderated by the software Polar Precision Performance™SW (version 4.01.029), followed by the manual one, to eliminate premature ectopic beats and artifacts. To evaluate the non-linear behavior of HRV, symbolic analysis and SE were used. The symbolic analysis evaluation is based on the quantification of the information carried in a temporal series, on the transformation of the previously selected R-R intervals into integers from zero to six, fromwhich the symbolic patterns (three-symbol sequence) are constructed. All possible patterns were grouped without loss into four families, according to the number and type of variations between subsequent symbols, 9 namely: (1) patterns, without variation (0V), that is, three equal symbols, for instance, "4,4,4"; (2) patterns with one variation (1V), that is, two subsequent equal symbols and a different one, for instance, "4,2,2"; (3) patterns with two similar variations (2LV), that is, the three symbols form an upward or downward slope, for instance, "1,3,4" or "5,4,2"; (4) patterns with two different variations (2ULV), that is, the three symbols form a peak or a valley, for instance, "3,5,3" or "4,1,2". The pattern related to the sympathetic branch performance is represented by the 0V family, and the performance of the parasympathetic branch is represented by the 2LV and 2ULV patterns. The joint performance of the ANS branches is represented by the 1V family. 9 The frequencies of occurrence of these families (0V%, 1V%, 2LV% and 2ULV%) were evaluated in this study. To calculate these indexes, we counted the number of times a pattern, which belonged to a specific family, was found, using a specific non-linear analysis software. 9 Another variable calculated in the same software was SE, which represents the complexity of pattern distribution. SE was used to quantify the complexity/regularity of heart rate fluctuations. Based on Shannon's framework, entropy is the measure of information of a given message – a message with low entropy/information is characterized by repetition. 9 For the analysis of the linear HRV indexes, the RMSSD indexes and the standard deviation of the mean R-R intervals (SDNN) were used in the time domain. 4 As for the frequency domain, the low (LF: 0.04 to 015 Hz) and high frequency (HF: 0.15 to 0.40 Hz) spectral components, in milliseconds squared and normalized units (n.u.), and the ratio of these components (LF/HF), were used. The spectral analysis was calculated using the Fast Fourier Transform algorithm. 4 The HRV analysis software (Kubios, Biosignal Analysis and Medical Image Group, Department of Physics, University of Kuopio, Finland) was used to calculate these indexes. 17 Data analysis In order to characterize the assessed volunteers, the descriptive statistical method was used, and the results are shown in absolute numbers, means and standard deviation for data with normal distribution (height, RMSSD, LF n.u., HF n.u., 2LV and SE) and median and interquartile range for those with non-normal distribution (age, body mass, BMI, SDNN, LF ms 2 , Hf ms 2 , LF/HF, 0V, 1V and 2ULV). For comparison between the groups (control and DM1), the normality of the data was initially determined using the Shapiro-Wilk test. When the normal distribution was accepted, the Student’s t test for independent groups was applied and, in the cases where the normal distribution was not accepted, the Mann-Whitney test was applied. The effect size of the difference between the comparisons was analyzed by Cohen’s d and values above 0.80 were adopted with high magnitude. 18 In order to verify the association and agreement between the indexes, correlations were carried out between linear and non-linear HRV indexes and, for this purpose, Pearson’s correlation was applied to the data with normal distribution, or Spearman’s correlation, for the ones that did not accept this distribution. Correlation values of r from 0.7 to 1 were considered strong, 0.4 to 0.6 were considered moderate and values of 0.1 to 0.3 were considered weak. The Intraclass Correlation Coefficient (ICC) was calculated. The statistical significance was set at 5% and the Confidence Interval at 95% (95% CI). Data analysis was performed using the software MiniTab version 13.20 (Minitab, Pa, United States) and Statistical Package for Social Sciences (SPSS), version 15.0 (SPSS Inc., Chicago, IL, United States). Results Of the 88 assessed volunteers, six showed errors in the R-R interval series greater than 5% and were excluded. Therefore, we analyzed the data of 39 young adults with DM1 (20 females) and 43 healthy young individuals (22 females), whose characteristics can be seen in table 1. The DM1 group had higher values of body mass and BMI (p < 0.05). Of the young individuals with DM1, 38.46% used other drugs in addition to insulin. Medications for the control of blood pressure (12.82%) and cholesterol (7.69%) were also used by these individuals. Moreover, 20.51% used drugs for thyroid disorders, 12.82% used contraceptives and 20.51% used medications for several diseases, such as rhinitis, diabetic polyneuropathy, peripheral neuropathy and epilepsy. Table 2 shows the linear index values in the HRV time domain for both groups. Significantly lower values were observed in the DM1 group when compared to the control group for both indexes (SDNN and RMSSD). The effect of the difference between the groups was considered high, as demonstrated by the obtained value of d = 1.210 and d = 1.203 for the SDNN and RMSSD indexes, respectively. 96