Lung ultrasounds were performed by one operator, specially trained to recognize and interpret B-lines. Lung ultrasound was performed on patient in a supine position using a Doppler portable ultrasound tool with a vascular probe of 6 - 12 MHz (BK Medical Mini Focus^®) for the detection of B-lines. Comet-tails or “B-lines” are defined as hyperechoic reflections which originate only from and travel roughly perpendicular to the pleural line of the lung. They have a narrow base and form a ray spreading away from the transducer towards the bottom of the screen (similar to a rocket at lift-up) and synchronously move with lung respiration [23] [24] . Areas explored included anterior and lateral regions of the two hemithorax, from 2^nd to 4^th intercostal space, and from 1^st to 5^th intercostal space in the right side of the chest. For each intercostal space, B-lines were detected in 4 different sites: para-sternal, midclavicular, anterior axillary and mid axillary areas, giving 28 positions per test [25] [26] . For all the explored sites, B-lines ranged from 0 to 10. Number 0 indicated the absence of B-lines whereas number 10 stood for complete blank screen. All the B-lines detected on various sites explored produced a score, the score of B-lines, indicating the severity of lung congestion [25] [26] . Patients were grouped according to these scores and to the following levels of lung congestion as previously described in other studies: No congestion ≤ 5 B-lines, mild congestion: 6 - 15, moderate congestion: 16 - 30, severe congestion: >30 [9] [22] [26] [27] . Figure 1 shows examples of B-lines.

Clinical examination was performed before the dialysis and 30 minutes after the session. All patients were evaluated during the second dialysis session of the week. Hemodynamic and anthropometric parameters such as blood pressure, heart rate, weight and height were taken. Clinical examination substantially consisted in identifying symptoms and signs of hyperhydration or fluid overload and of dehydration or hypovolemia. Each functional or physical sign related to hyperhydration such as peripheral oedema, dyspnea, high blood pressure, jugular veins turgor, ascites [6] [24] [28] [29] [30] was given a positive figure whereas negative figures were attributed to signs indicating hypovolemia such as dizziness, asthenia, cramps, low blood pressure [6] [24] [28] , resulting to a clinical score as interpreted in Table 1.

This clinical score before and after dialysis were compared, to determine the impact of dialysis on clinical signs assessing the hydration status.

Data were recorded and treated with Excel 2013, then exported to BM-Statistical Package for Social Sciences (IBM-SPSS) 22.0 for statistical analysis. Tables and figures were obtained using Excel 2013.

Data were presented as mean and standard deviation, median and interquartile range, or frequency and percentage, as appropriate. We used the Shapiro Wilk test to determine the normality of our quantitative variables’ distribution.

For qualitative variables, we used the Chi-square test to compare groups of patients or the Fischer test when the expected number of patients was below 5, and the Mc Nemar test was used to compare variables.

For quantitative variables, the student test-t was used for independent samples and the Wilcoxon signed-rank test, where appropriate. Correlations between variables were assessed using Spearman’s rank correlation coefficient. The relationship between the B-lines before and after dialysis was measured using the intra-class correlation coefficient. The significance threshold was established for p < 0.05.

After dialysis, it was observed that the number of dyspnoeic patients significantly reduced but more patients complained of cramps (Table 2). Up to 42.5% of patients proved to have fluid overload before dialysis; whereas 16.1% showed a normal clinical hydration status (Table 3).

Levels of systolic blood pressure and weight decreased considerably after dialysis. Blood pressure decreased from 164.74 (±26.50) to 158.48 (±27.89) mmHg and weight loss from 74.76 (±11.4 to 71.38 (±11.34) kg. It was also observed that their clinical score significantly dropped after dialysis (p < 0.001).

B-lines number median reduced considerably after dialysis (p < 0.001), as described in Table 4.

Values are given in frequency and percentage in brackets; significant p values are in bold.

Values are given in frequency and percentages in brackets.

Values are given in frequency and percentage in brackets; lung congestion < 6 B-lines = no congestion, [6 - 15] = mild congestion, [16 - 30] = moderate congestion, >30 = severe congestion.

The clinical factors associated with the presence of B-lines before dialysis were dyspnoea and jugular veins turgor. This association decreased after dialysis. Indeed the number of dyspnoeic patients increased with the severity of lung congestion assessed by the number of B-lines (Figure 2). It was the same observation with the jugular vein’s turgor (Figure 3). This number was very low after dialysis.

We assessed the number of B lines before and after dialysis, the decrease in the number of B lines after dialysis and the various measured clinical parameters.

Clinical parameters associated with the number of B-lines before dialysis:

There was a big correlation between the number of B lines before and after dialysis (r = 0.805, p < 0.001), measured by the interclass correlation coefficient. The number of B-lines was also associated with the clinical score before dialysis (r = 0.549; p = 0.001).

Clinical and ultrasonographic parameters associated with the decrease in the number of B lines:

Table 5 shows that the decrease in the number of B-lines after dialysis was strongly associated with the number of B lines before dialysis (r = 0.862; p < 0.001).

Clinical and sonographical parameters associated with the number of B-lines before dialysis:

Findings in Table 6 show that the number of B-lines after dialysis was strongly correlated with that of the presence of B-lines before dialysis (r = 0.935; p < 0.001) and with the reduction of B-lines after dialysis (r = 0.672; p < 0.001).

In this study, we observed that there is a significant decrease in the number of B-lines after dialysis, as described in many previous studies [9] [20] [24] [31] . B-lines are associated with the accumulation of water during the interdialytic period and reflect the variation in the level of extravascular lung water which occurs during dialysis [32] . In their study, Khamis et al. also showed a significant decrease of number of B-lines post-dialysis, especially for the group of patient with interdialytic hypertension (from 10 to 4 B-lines) [20] . Noble et al. [33] and Mallamaci et al. [22] proved that the number of B-lines rapidly and in a reliable manner reflects the extracellular fluid variation, with an average duration of 4 minutes for the change on the screen [22] . Whereas inferior vena cava ultrasound can only be performed two hours after dialysis because of the equilibration time between interstitial and intravascular compartments [22] [34] [35] . Equally, multifrequency bioimpedance should be done 30 minutes to two hours after dialysis [36] [37] .

After dialysis, we observed a significant drop in the systolic blood pressure, though it was low compared to studies carried out in Europe [9] [22] [31] [38] . This could be explained by the fact that the patients described within this study presented with a higher blood pressure probably related to the reduction of weekly dialysis sessions and the irregular intake of antihypertensive drugs. There was also a significant drop in the clinical score showing that it could reflect the variation in hydration status after dialysis.

A correlation between dyspnea and the presence of B-lines was observed before dialysis; a similar correlation was also observed by Siriopol et al. [9] and Mallamaci et al. [22] . The role of fluid overload in the development of pulmonary oedema was highlighted by the significant decrease in the number of dyspnoeic patients after dialysis [9] [22] .

We also observed a correlation between the presence of B-lines in pre-and post-dialysis and jugular veins turgor (p = 0.005). This correlation reduced after dialysis (p = 0.022). Furthermore, jugular veins turgor was more sensitive than dyspnoea in predicting pulmonary oedema. Indeed, jugular veins turgor could be felt with mild lung congestion (Figure 2). This correlation with jugular veins turgidity was not found in the literature, thus the specificity of our study. Moreover, dyspnea and jugular veins turgor were observed only after a minimum of 6 B-lines, indicating that lung ultrasound could help detect the formation of pulmonary oedema at the preclinical stage, also shown by Mallamaci et al. [22] and Allinovi et al. [28] . Therefore, it is a sensitive indicator of the efficiency of ultrafiltration given the regression of dyspnea post-dialysis (Figure 3).

We did not find any significant difference in peripheral oedema at pre- and post-dialysis. This explains the non-specificity of this clinical sign in the assessment of hyperhydration as described by Torino et al. [39] .

We could also establish a clinical score based on patients’ physical and functional signs, and this score was positively correlated with the number of B-lines before dialysis (r = 0.49, p = 0.005). These findings are similar to those presented by Allinovi et al. [28] who found a linear correlation between B-lines number and the clinical assessed level of fluid overload.

Despite significant weight loss post-dialysis (−3.38 kg, p < 0.001), there was no correlation between the number of B-lines after dialysis and weight variation. This is similar to findings by Siriopol et al. [9] , Khamis et al. [20] and Mallamaci et al. [22] . Whereas Vitturi et al. [24] obtained a relatively low number of B-lines, and they described a high correlation between the decrease in the number of B-lines and weight loss after dialysis. However, their target population was made up only of asymptomatic patients with clinically determined dry weight [24] . Therefore, the absence of a correlation between weight variation and the reduction of the number of B-lines in our study could probably be due to the fact that patients were not at their dry weight, thus the relatively high number of B-lines; and secondly from the high number of dyspnoeic patients.

Studies carried out by Mallamaci et al. [22] showed a high correlation between the number of B-lines, the reduction of the number of B-lines and the cardiographic parameters, particularly the ejection fraction of left ventricle. According to them, the presence of B-lines essentially depends on the fraction of systolic ejection rather than on the fluid overload. In the same light, Siriopol et al. [38] found that lung ultrasound could produce the same information as echocardiography; and so, could improve treatment and prognosis of chronic HD patients, since it’s easier, more accessible and cheaper than echocardiography.

In the light of these analyses, lung ultrasound could be a good alternative for an objective and precise assessment of HD patients’ hydration status. It could contribute to a reliable weight loss determination, as described by other researchers [24] [30] [34] . It is a non-invasive test, easy to perform by non-specialists [22] [28] , and could easily be used in Africa and in Cameroon, particularly given the availability of ultrasound machines in our hospitals and in some nephrology units. Compared to BIS or inferior vena cava diameter measurement which are costly, this test simply requires standard equipment [23] [31] . Its intra- and inter-observator reproducibility is good and higher than that of inferior vena cava ultrasound measurement [22] [24] and with a good inter probe reproducibility [22] , making it easy to perform this test even with a probe for kidney evaluation (3.5 MHZ). Moreover, studies have shown that determination of extravascular lung water using a lung ultrasound machine was highly associated with the mortality and the occurrence of cardiovascular difficulties in chronic HD patients [9] [27] , thus indicating its role in assessing haemodialysis prognosis. This ultrasound technique has the advantage that it can be performed immediately after a dialysis session on different categories of HD patients (young children, patients with metallic implants, amputees, etc.). One of the major benefits of lung ultrasound is its superior correlation with every other hydration status assessment techniques (bioimpedance, sonographic measurement of inferior vena cava, clinical methods) [24] [28] [31] .

From the analysis above, the integration of lung ultrasound in the follow up of chronic HD patients is recommended for an objective assessment of their hydration status and the precise determination of their dry weight.