Abstract
Background and Aims:
Fluid balance for the surgical patient has been proven very important for the postoperative outcome and development of complications. The aim of this study was to evaluate, for the first time in modern times, the accordance between nurse-based fluid charting (cumulated fluid balance) and body weight change for general surgical patients.
Material and Methods:
This was a descriptive study with prospectively collected data from two clinical randomized multicenter trials. A total of 113 patients from American Society of Anesthesiology group I–III undergoing elective colorectal surgery were included. Cumulated fluid balance and body weight change were charted preoperatively and daily at the same time during a postoperative period of 6 days. Differences were calculated by subtracting cumulated fluid balance from body weight change (1 g = 1 mL), and agreement was assessed by making Bland–Altman plots as well as Pearson correlations.
Results:
From day 1 to 4, the mean difference between cumulated fluid balance and body weight change was below 0.4 kg/L. On day 5 and 6, the discrepancies increased with mean differences of, respectively, 1.2 kg/L (p < 0.002*) and 2 kg/L (p < 0.0001*). Bland–Altman plots showed increasingly poor agreement for all postoperative days with wide limits of agreement, ranging from more than 6 kg/L to almost 10 kg/L. Pearson correlations were moderate to strong at all times ranging from 0.437 (day 1) to 0.758 (day 4).
Conclusions:
The accordance between cumulated fluid balance and body weight change for colorectal surgical patients is relatively good for the first four postoperative days, however, with large uncertainty, whereas on the fifth and sixth postoperative day, the discrepancy is statistically and clinically significant. The fluid chart cannot stand alone in interpretation of the patient’s fluid balance; body weight and clinical judgment is indispensable.
Keywords
Introduction
Fluid therapy in patients undergoing colorectal surgery has been heavily debated in the last years, and the optimal intra- and postoperative fluid administration is the subject of controversy. Too little fluid in patients undergoing colorectal surgery might lead to postoperative complications, circulatory collapse, and death (1). However, fluid overload also causes harm such as cardiopulmonary complications, poor tissue healing, and increased risk of death after surgery (2–4).
For this reason, it is important to evaluate the accuracy of the tools with which we measure fluid balance (FB) in daily clinical practice.
Different parameters are used to assess volume status for the surgical patients. Inputs and outputs are recorded to calculate the FB and, thus, quantitative as well as qualitative aspects of the fluid therapy are assessed. In the best of circumstances, the body weight (BW) is measured repeatedly and physiological parameters such as blood pressure (BP), heart rate (HR), and arterial oxygen saturation are monitored (1). Analysis of urine osmolality or hemoglobin (or hematocrit) might also help to determine the best fluid therapy to be given (5).
Documentation of inputs and outputs may be challenging in a busy surgical department. Perspiration is always based on an estimate and risk of incorrect measurements as well as errors of registration and calculation may occur. The reliability of fluid charts has been questioned by various studies comparing daily and cumulated FB with BW changes (6–10). In these studies, it was shown that FB did not reflect the change in BW very well, and by consequence, the fluctuations in total body water. Perren et al. (7) concluded that clinical decision-making should be based on more objective techniques (i.e. standardized measurement of BW or whole body impedance changes) when it comes to fluid therapy. These trials, however, examined patients admitted to the intensive care unit (ICU) and patients undergoing cardiac surgery. No trials were found for patients undergoing colorectal surgery since the thesis of Francis. D. Moore in 1952 (11).
It is clear that, on one hand, fluid therapy is important for the outcome following surgery, and, on the other, it is difficult to measure correctly. Compared to the ICUs, the daily fluid charting in the surgical department is challenged by having a smaller nursing staff and patients that can usually eat, drink, and visit the toilet themselves. Furthermore, losses such as evaporation from the surgical wound and insensible perspiration in patients with fever may be very difficult to estimate. Especially for extended admissions, errors add up and cumulated FB may become more and more incorrect.
In this study, it was hypothesized that the daily clinical practice of nurse-based fluid charting would not agree with BW change in a sample of surgical patients during the postoperative period. The relationship was evaluated by investigating agreement between recorded cumulated FB (defined as the sum of all daily FBs) and the BW change by using the formula, 1 mL = 1 g.
Material and Methods
In this descriptive study, data from two clinical randomized multicenter trials were used, and both trials were approved by the relevant scientific ethical committees (2, 12). In the original trials, adult patients admitted for elective colorectal surgery, categorized as American Society of Anesthesiologists (ASA) group I–III and could give informed consent were considered eligible for inclusion. Patients were excluded if they drank more than 5 drinks a day (or 35/week), had mental disorders, diabetes mellitus, renal insufficiency, disseminated or secondary cancer, diseases hindering epidural analgesia, or were pregnant or lactating women. Data originating from two Danish high-volume surgical centers were included because FB and BW were both recorded at 08:00 a.m. every day (other centers were not included because the FB was calculated at different times of day). Patients with indispensable missing data such as no preoperative BW, no postoperative BW, and no registered weight of the colonic preparation were excluded.
Weight and fluid registration: All patients had their weight measured and recorded in the morning before the operation and every subsequent morning as close to 08:00 a.m. as possible in daily clinical practice on a surgical ward. Registration continued until discharge or until the sixth postoperative day. To avoid inaccuracy between different scales, each patient was weighed on the same scale every day. The weight of the removed colon or rectum was recorded and corrected for by subtracting this weight from the patient’s initial preoperative total BW.
All fluid inputs and outputs were recorded from beginning of fasting (preoperatively) to the sixth postoperative day or until discharge. Fluid losses included urine, stoma output, secretion from drains, and aspiration. Solid feces and diarrhea were not recorded. The drains, stomas, urine bags, and so on were emptied as close as possible to 08:00 a.m. in daily practice on a surgical ward. For all patients, the insensible perspiration was calculated as 10 mL/kg/day without adjustments for time of surgery or changes in room- or body temperature (1). Sweating associated with fever was not recorded. Fluid intakes included intravenous (IV) fluid, oral fluid, enteral nutrition by tube, medication, or blood products. No specific registration of solid food was made, however, the possibility to record “other enteral intake” was present. An estimation of remaining volume in IV fluid bags was done as close to 08:00 a.m. as possible.
Statistical Analysis
IBM SPSS Statistics 20 was used for data analysis. Previous trials included 32 (6), 147 (7), and 151 (9) patients in the analysis. The 113 patients included in this study, with prospectively collected data, were estimated sufficient based on similar studies. The data were tested for normality and when present parametric statistics (T-test and standard deviations (SDs)) were used. The significance level was considered as p ≤ 0.050. All p-values were two-sided—the difference between cumulated FB and BW was calculated by subtracting one from the other, by using the formula 1 g = 1 mL, and presenting the mean difference and variation (SD), as well as calculating the Pearson correlation coefficient. Agreement between the two variables for each day was assessed by the Bland and Altman method (13). The 95% limits of agreement were calculated as mean error ± 2SD. The number of patients decreased from day 1 to day 6 because of discharge or incomplete data. Data entry and calculations were done twice to reduce the risk of calculation errors.
Results
In the original trials, 291 patients were included (2, 12). Because FB and BW measures were evaluated at 08:00 a.m. at two hospitals, a total number of 130 patients from these hospitals were eligible. Of these, 17 patients were excluded because of missing data leaving 113 for data analysis. Six patients were excluded because the preoperative weight was missing, nine patients because the postoperative weight was missing, two patients because the weight of the preparation was missing. Regarding basic and surgical properties, the excluded patients did not differ from the included ones. Basic and surgical data are shown in Table 1.
Basic and surgical data.
ASA-score: American Society of Anesthesiologists Score; SD: standard deviation.
The mean BW change and the mean cumulated FB with SDs are shown in Fig 1. It is seen that there is a good accordance between the two from day 1–4. However, on the fifth and sixth postoperative days discrepancy increases between the two, that is, the cumulated FB is more positive than the BW change (p = 0.002 on day 5 and p ≤ 0.0001 on day 6). In addition, the figure shows that the BW increases postoperatively until the third postoperative day and decreases thereafter.
Average differences between cumulated FB (dark color) and BW change (bright color) for six postoperative days. FB is adjusted for with insensible perspiration as 10 mL/kg/day. Standard deviations for means are indicated.
BW change minus cumulated FB is shown in Table 2 with SDs. The mean error is very small and less than 100 mL the first two postoperative days. The error increases thereafter and becomes significantly different on postoperative day 5 and 6, with SDs of the mean error of more than 2000 g/mL on day 5 and 6.
Body weight change minus cumulated fluid balance. The mean error is small on the first two postoperative days and increases thereafter and becomes significant on postoperative day 5 and 6.
df: degrees of freedom.
Indicates significant difference.
Bland-Altman plots were made for all six postoperative days as shown in Fig 2. In addition to the growing mean error (bias), we found poor agreement between the cumulated FB and BW change as suggested by the 95% limits of agreement that are very wide, ranging from 6361 g/mL (day 2) to 9998 g/mL (day 5).
Bland-Altman plots. Y-axis shows the difference between BW change and cumulated FB (g/mL). X-axis shows the (BW change + fluid balance)/2 (g/mL). Dotted lines represent the 95% limits of agreement.
The Pearson’s correlation coefficients with scatter plots are shown in Fig 3. There is a positive linear relationship between cumulated FB and BW. The weakest correlation is calculated within day 1: r = 0.437**. The remaining correlations are stronger with some variation: day 2: r = 0.745**; day 3: r = 0.587**; day 4: r = 0.758**; day 5: r = 0.665**; and day 6: r = 0.646**. All correlations are significant with p ≤0.010. The ideal correlation (Y = X + 0) is shown as a dotted line, which shows that the cumulated FB is consequently overestimating the BW change overall. However, on day 1 and day 2 only, respectively, 44% and 47% of the patients had cumulated FB greater than BW change. On the subsequent 4 days cumulated FB was greater than BW change for, respectively, 57%, 54%, 65%, and 71% of the remaining patients.
Correlation plots. Y-axis shows BW change preoperatively to the following 6 days (g). X-axis shows the net cumulated fluid balance preoperatively to the following 6 days (mL). Dotted line marks the ideal correlation with Y = X + 0.
Discussion
To our knowledge, the accuracy between BW and FB has not been studied for general surgical patients since the thesis of Francis D. Moore in 1952 (11). The data were collected prospectively in two large volume surgical departments with inclusion of a relatively large amount of patients. Weight and FB were charted postoperatively, and daily at the same time, making it possible to evaluate the day-to-day development of, respectively, BW change and FB and the difference between the two. Previous studies have included patients admitted to the ICU (7–9) or cardiac surgical patients (6), see below.
General surgery has several challenges that are not present for studies of ICU patients or cardiac surgical patients. First, a preparation is removed during surgery, clearly influencing the BW. Second, the evaporation from the surgical wound is difficult to measure, but its magnitude has been severely overestimated for many years, while it in fact is very small (14). Moreover, the evaporative loss for patients operated laparoscopically is completely unknown. However, the insensible water loss from the lungs are eliminated during surgery, because the patients are ventilated with moist air, thereby decreasing the insensible perspiration. We chose not to attempt to correct for any of these operative changes in fluid loss. Third, abdominal operated patients have a postoperative intestinal paralysis, causing a positive FB and a weight gain not seen in patients undergoing cardiac surgery. As seen from Fig 1, the increase in BW from day 1 to 3 with a subsequent decrease the following days most likely reflects this. The illogical increase in BW on day 6 most probably reflects small numbers, however, the patients remaining on the sixth postoperative day are characterized by having prolonged intestinal paralysis, surgery of the rectum, or having a stoma (training for handling the stoma).
Our results are in agreement with the findings of previous studies investigating patients from ICU’s and patients undergoing cardiac surgery (6–9). Roos et al. (8) found that neither bioelectrical impedance measurements nor carefully calculated FBs corrected for insensible losses could be predictive of, respectively, total body water or BW changes in critically ill patients from an ICU. First, because the insensible water loss could not be predicted accurately. Second, because FB did not take into account the composition of body mass such as muscle and fat tissues. In a study by Eastwood et al. (6), with patients undergoing cardiac surgery, they found only 3 (9.75%) in 32 patients with ±250 mL agreement between BW change and FB. In 19 (59.4%) patients, FB underestimated weight gain, and in 10 (31.2%) patients, FB overestimated weight gain, which is unlike our study where FB overestimated weight gain for the majority of patients after day 2. Perren et al. (7) studied 147 patients in an ICU and found, like in our study, poor agreement between cumulated FB and BW; however, with a correlation for adjusted cumulated FB and BW changes of r = 0.845 (R2 = 0.714), thus a stronger correlation was presented in our study. Especially, for day 1 (r = 0.437**), our correlation was weak. This may be caused by the different operative losses not accounted for, but may be even more importantly by problems in communication and registration between the three different wards a surgical patient is transported to on the first day (the surgical ward, the operation room (OR), the recovery room, and back to the surgical ward). Finally, a recent study by Schneider et al. (9) examined the correlation between BW (using electronic bed weights) and FB on two consecutive days in an ICU and found a weak Pearson correlation coefficient of r = 0.34 (95% confidence interval, 0.26–0.42, p < 0.001) and large 95% limits of agreement (almost 12 kg). One could hypothesize that immobilized ICU patients with urinary catheters, drains, possibly stomas or rectal tubes, and bags for defecation, tubes for enteral nutrition, IV lines, and so on would have better correlation between BW change and FB simply because it would be easier to measure inputs and outputs. However, the results of this group of patients were not consistently better than the results of this study.
However, our study has weaknesses. Like in previous studies, patients had to be excluded because of missing data. Registering a complete fluid chart in a surgical department is difficult because patients may forget to collect urine or to register the drinking of a glass of water. Furthermore, calculation errors, including risk of duplicating input or output, are present in our trial as well as in the previous trials (6, 7, 9). Missing data may impose selection biases. For example, it is possible that the most ill patients were unable to get out of bed to get weighed, or, on the other hand, that the most fit went to the bathroom on their own and forgot to register loss. In addition, the insensible fluid loss is based on an estimate and calculated to be 10 mL/kg/day. This is in our part of the world a fair estimation but, of course, not completely accurate.
Solid food was not recorded in our data adding a general error to the BW. However, liquid food was registered, that is, nutrition by tube, protein drinks, soups, and so on. We are aware that the discrepancy between solid and fluid food is not clear, for example, yoghurt was not recorded as fluid despite large water content. This is not accurate in a strict scientific way (we do not place patients in a calorimeter), but this is how fluid is recorded in daily clinical practice and, therefore, the data that the surgeon is given to act upon.
Moreover, the surgical trauma with postoperative pain and stress response cause muscle wasting, especially in elderly, immobilized and not sufficiently nourished patients (15). As seen from Fig 1, the weight of the patient decreases after day 3 as well as discrepancy between BW change and cumulated FB increases. This may, to minor a extent, be explained by the fact that surgical patients become catabolic after the surgical trauma. However, this is counteracted by the infusion of IV fluids with densities above 1 kg/L, see below. Finally, as seen from Fig 1, the mean postoperative BW does not become smaller than the preoperative BW in our study.
Our results are based on the assumption that 1 mL = 1 g. This is only approximately true since IV fluid contains various concentrations of electrolytes making their density slightly heavier than water. The weight of 1 L normal saline (NS) is 1.005 kg/L, 1 L of 5% glucose is 1.016 kg/L, 1 L of hydroxyethyl starch (HAES) 6% is 1.027 kg/L, 1 L of erythrocytes in saline-adenine-glucose-mannitol (SAG-M) weighs 1.056 kg/L, and so on (16). Indeed, this introduces a small systematic error, and ideally, additional 5 (NS) to 56 g/L (SAG-M) should be added to the weight of the patient after infusion depending on the quality of fluid. However, given the uncertainties in the calculation of perspiration and the other above mentioned difficulties, correction for such small errors is without meaning.
Correct assessment of FB is essential to supply the surgical patients with the right amount of fluid both during and after surgery, and it is important for patient outcome following surgery. Hypovolemia leads to impaired oxygen transport, failure in organ perfusion, and death. Several studies, however, clearly show that fluid overload leads to poor patient outcome as well. Excessive amounts of intraoperative fluid causes decreased pulmonary function (17), reduced gut motility (3, 18), and postoperative complications and increased morbidity in both vascular surgery (19, 20), pneumonectomy (21, 22), and liver transplantation (23). Furthermore, intraoperative fluid “restriction” predicts shorter length of stay in ICU after, respectively, spine (24) or vascular (25) surgery and in the gastrointestinal surgery restriction or “zero-balance” fluid therapy reduces postoperative complications (2, 3, 26, 27).
Our results suggest that neither the fluid chart nor the BW can stand alone in the interpretation of the patient’s FB. A daily weight measure is indispensable to measure the quantity of fluid changes, while the fluid chart is needed for the quality of losses. However, fluid may be sequestrated in extra vascular spaces such as the intestinal lumen. Therefore, clinical parameters like BP, HR, and urinary output are obligate as measures of hypovolemia. More technical approaches as monitoring with an esophageal Doppler, a pulmonary artery catheter or PiCCO monitoring (28) provides information about the cardiac output but is not feasible in a surgical ward.
In conclusion, the accordance between cumulated FB and BW change for colorectal surgical patients is relatively good the first four postoperative days, however, with large uncertainty, whereas on the fifth and sixth postoperative day, the discrepancy is statistically and clinically significant. The fluid chart alone is not sufficiently accurate in the estimation of FB and clinical judgment as well as BW must be taken into account to ensure the best possible fluid therapy for the patient.
Footnotes
Acknowledgements
The authors thank all the surgeons, anesthetists, and nurses who collected data for the two original trials. Without their dedication and thorough efforts, this article would not exist.
Declaration of Conflicting of Interests
The authors declare that there are no conflicts of interest.
Funding
This study was funded by Aase and Einar Danielsens Fund; Eastern Danish Health Science Research Forum; Hans and Nora Burchards Memorial Fund; The Danish Hospital Foundation for Medical Research Region Copenhagen, Faeroe Islands and Greenland; The Danish Medical Association Research Fund; Olga Bryde Nielsens Fund; Inge and Jørgen Larsens Memorial Fund; and Grosserer Valdemar Foersom and wife Thyra Foersom Fund.