Patients

This comparative study was performed in Dialy-Ser, a renal unit located in Bogotá, Colombia. Data were collected from 154 patients on hemodialysis during the program, between August 2011 and August 2012. 77 patients were defined as incident patients, because their hemodialysis treatment had begun less than three months ago. 77 patients were defined as prevalent patients, because they had been on dialysis therapy for three months or more.

The study included 154 patients (43.5% women and 56.5% men), older than 18 years. The quality of life of the subjects was evaluated using the Medical Outcome Study 36-Item Short-Form Survey (SF-36), version validated in Spanish. Eight areas of health were included: physical functioning (PF), role participation with physical health problems - physical role (RP), corporal pain (BP), general health (GH), vitality (VT), social functioning (SF), role participation with emotional health problems (emotional role) (RE) and mental health (MH).

In addition, to ensure standardized conditions, summary measures of physical and mental component (PCS and MCS) were taken into account. Three scales, PF, RP and BP, are related to the physical component, since they contribute more to the score of the Physical Component measure (PCS). The mental component is most highly related to the scales MH, RE, and SF, which contribute most to the score of the summary measure of mental components (MCS). Finally, three scales have worth mentioning with both components, namely: VT, GH, SF.

Questionnaires were provided, answered by the participants, during the regular hemodialysis sessions, with the help and explanation of the nurse trained in CKD. All patients filled out the questionnaires completely. Through these questionnaires, medical records and interviews with patients, demographic and medical data were obtained, including age, sex, marital status, education, occupation, vascular access, BMI, comorbidities. Likewise, laboratory data, including hemoglobin, albumin, potassium, calcium, phosphorus, glucose and parathyroid hormone PTH, were obtained from fasting blood samples taken during the week of RRT initiation.

Ethical aspects

Informed consent was required for inscription, after patients received verbal and written information about the study. Informed consent was received from all individual participants included in the study. Ethical permission for the study was obtained from the scientific committee of the renal unit. All the procedures performed in studies with human participants were in accordance with the ethical standards of the institutional research committee and with the Declaration of Helsinki of 1964, its subsequent amendments or comparable ethical standards.

Statistical analysis

In our study, the CART model was used to identify the variables that most adequately discriminate between incident and prevalent patients. Classification and regression trees (CART) are widely used in predictive analysis, since they explain non-linear effects, offer quick solutions for the hidden and complex substructure and provide unbiased information, statistically significant, high-dimensional analysis, apparently not related to the data. CART is a non-parametric method, which uses mixtures of categorical (nominal and ordinal) and continuous data in the result, as well as predictors. It has the ability to control the effect of hidden interactions in a better way, compared to alternative methods. In addition, the results provided by tree-based analyzes are interpreted, visually and logically, more easily.

In order to evaluate the possible differences in SF-36 variables between incident and prevalent patients, the MANOVA model was used and the null hypothesis of equality of means between the SF-36 variables was tested. Subsequently, Marginal ANOVA was performed for each variable, with the purpose of detecting the differences observed in detail. In addition, a PCA was performed to identify possible groups of patients associated with certain physical or mental characteristics, which are distinguished by SF-36 variables and which are identified by the contributions of each variable to the main components.

All statistical analyzes were performed in R version 3.0.2.

Quality of Life

Eight SF-36 quality of life variables were evaluated by MANOVA multivariate analysis. Significant differences were found between the two groups (p = 0.028): except for body pain and emotional problems, in all other variables it is observed that the incident patients have less value in the quality of life than the prevalent patients (figure 2). Similarly, the analysis for PCS and MCS found significant differences (p = 0.0027): the incident patients tend to have lower quality of life scores than the prevalent ones. (Figure 3).

Figure 2 Box plots. Comparison of quality of life between incident patients and prevalent in the SF-36 items. 

Figure 3 Ellipses. Bivariate 95% confidence regions for incident patients (0) green and prevalent patients (1) red. 

Principal component analysis

It was found that the first component (with an explanatory power of 46%) is positively associated with all the quality of life variables that indicate that patients tend to have a high or low status in the set of all the variables (figure 4). The second component includes MH and GH. These are moderately associated and negatively associated with RP and VT. They reveal that some patients with low scores on the scale of limitation, caused by physical problems and vitality, are not significantly affected in the scales of mental health and perception of the disease, since a great majority of patients have positive values in the component main 2 (figure 5). On the other hand, we found a significant group of prevalent patients with good scores on the physical component scales.

Figure 4 Dispersion plot for the two main components. Incident patients (0) green, prevalent patients (1) red. 

Figure 5 Correlation circle for the SF-36 variables. Complementary variables to MCS and PCS. 

According to the SF-36, in our population there is a high relation between the sum of the mental component and the variables MH, RE, GH and, to a lesser degree, BP and SF, while the sum of the physical component is highly associated with PF, VT. In addition, it is observed that, although the physical pain is involved with the construction of the two indices, it was found more strongly associated with the mental component than with the physical one.