Impact of a computer-assisted decision support system (CDSS) on nutrition management in critically ill hematology patients: the NUTCHOCO study (nutritional care in hematology oncologic patients and critical outcome)

HM patients increasingly require admission to the intensive care unit (ICU) for life-threatening events related to the malignancy and/or treatments [1, 2]. The prognosis of HM patients has improved considerably over time, in fact mortality in HM patients septic shock has fallen by 30% [3], while in patients receiving invasive mechanical ventilation (IMV) improvement has resulted in a sharp drop in mortality, from nearly 90% to about 40% [4]. Thus, critically ill HM patients require an extensive diagnostic workup [5] and the optimal use of available treatments [6]. Nutritional support is crucial in critically ill patients. Guidelines recommend early enteral feeding supplying 25 kcal/kg per day during the acute phase of critical illness [7, 8] with however a low level of evidence. For HM patients treated with myeloablative conditioning for allogeneic stem cell transplantation, early nutritional intervention has proven a benefit on outcome and mortality outside the ICU [9]. There are no data regarding nutrition strategy for HM patients during ICU stay. As most of ICU therapies, nutritional support has become more complex requiring tight supervision and monitoring. Thus, computer-assisted decision support systems (CDSSs), designed to implement rational care strategies based on guidelines and pathways, have been developed and studied over the past several years [10]. For this study, we hypothesized that implementation of a CDSS, designed to actively supervise a nutritional intervention by a multidisciplinary team, would be able to increase guidelines adherence and outcomes.

We report herein, in this before/after study, on 275 critically ill hematology patients admitted to the ICU over 5-year period. Energy and protein intakes were delivered using standard protocol in the 147 patients (53%) of the ‘before group’ and using a CDSS in order to reach every day predefined caloric and protein targets accordingly to the catabolic or anabolic status in the 128 patients (47%) of the ‘after group.’ We showed that the use of CDSS allows to relatively increase the rate of days in compliance with caloric and protein targets in the ‘after group’ by more than 50% as compared with the ‘before group.’ Interestingly, compliance rates were low and only reached 30% after intervention. Hospital mortality, ICU-acquired infection, and hospital and ICU length of stay were similar in the two groups of patients. Exploratory analysis showed that hospital mortality was lower in the ‘after group’ for neutropenic and severely ill patients.

In critically ill patients, guidelines recommend the use of early EN during the acute phase of the disease in order to prevent muscle wasting, infections, delayed recovery, and mortality [7, 8]. However, early EN may induce vomiting and gut ischemia during septic shock [21]. In hematology patients and in allogeneic HSCT recipients, guidelines recommend the use of early EN since it seems to be associated with better survival, less acute graft-versus-host disease (GVHD), and faster neutrophil recovery as compared with PN [11, 12, 22]. One striking finding from our study is that EN was difficult to realize in critically ill hematology patients, in fact the rate of EN days during ICU stay was low and about 13% and 25% for the before and after groups, respectively (Additional file 2: Table S2). The use of CDSS was however significantly helpful in this setting, particularly in the first 5 days of ICU admission during which only 4% of patients were nourished using EN before the use of CDSS and 26% after (Additional file 3: Table S3). We showed that the causes of EN failure could be frequently related to specific hematologic clinical patterns such as colitis or mucositis (Fig. 2). For these reasons, PN seemed to be the more appropriate feeding route as PN represented about 50% (‘before group’) of nutrition intakes during ICU stay and significantly increased to 63% (Additional file 2: Table S2). Accordingly, during the first 5 days of ICU admission, PN was the most frequent feeding route and the rate of patients nourished with PN significantly increased to 76% after CDSS use (Additional file 3: Table S3). Early nutrition appeared to be a crucial appointment in our study since 45% (‘after group’) and 58% (‘before group’) of the total caloric deficit was reached during the first 5 days of ICU admission (Additional file 1: Table S1). Thus, during the same period and after CDSS use, the number of patients who benefited from early nutrition (PN and/or EN) significantly increased as well as caloric intake (Additional file 3: Table S3). CDDS seemed to be an interesting tool to optimize energy intake and limit energy deficit in this setting [10, 23‐25]. Our results are in line with recent randomized clinical trials in which early PN was safe and as effective as enteral route which was associated with gastrointestinal complications [21, 26]. In NUTRIREA-2 study [21], caloric target was designed to match the course of acute disease in two steps (the acute and the recovery phases) [27], we used similar design for energy delivery. Despite this strategy, our patients were underfed due to the low compliance rate regarding energy target only reaching the value of 30% even after CDSS use. Similar results were found in the CALORIES study [26] in which the percentage of patients in whom energy target was met was about 30% for each day of the study period. Taken together, these results underline that energy intake might be better evaluated with daily rather than global assessment in which underfeeding might be underestimated. Indeed, in our study, the mean daily calorie intake in the ‘after group’ was comparable to recent data [28], about 1200 kcal/day (Additional file 1: Table S1), suggesting a moderate underfeeding contrasting with a more profound underfeeding when the assessment was detailed every day. Importantly, the low rate of compliance regarding caloric target was also due to overfeeding situations, thus 12% of patients were overfed at day 5 in the ‘after group’ (Additional file 1: Table S1). Finally, the failure to reach caloric target might be overestimated since the number of days in compliance with calories target was defined as days on which caloric delivery fitted the targets ± 15%. This definition appears to be very strict as compared with recent studies in which underfeeding was defined as caloric intake below 70% of the target [29].

There is growing evidence that protein intake might also affect outcomes [30, 31]; however, the best protein intake in critically ill patients is still unknown. Recent recommendations on protein intake for critically ill patients are only based on studies with a low level of evidence [32]. In the present study, we showed that the use of CDSS was associated with both a significant increase in protein delivery and decrease in protein deficit during ICU stay. Thus, mean daily protein intake after intervention was about 56 g/day (Additional file 1: Table S1) and similar to that recently described by Nicolo et al. [28]. As for energy intake, 47% (‘after group’) and 52% (‘before group’) of the total protein deficit was reached during the first 5 days of ICU admission (Additional file 1: Table S1) underlining once again the potent crucial role of early PN for those patients. Intensivists usually focus on caloric rather than protein goals, and the parenteral nutrition products we used have compositions that could not provide 1.5 g of protein/kg/day without overfeeding. Thus, prescribers needed to tradeoff between high calories or high proteins. Glutamine, added to total PN, could be a key to higher protein delivery without a caloric overshoot [33]. During the ‘after period,’ glutamine was available in our institution and thus integrated in the CDSS and could have allowed us to better reach our protein targets.

The hospital mortality was about 46% consistent with recent results [1]. For the overall population, CDSS use did not translate into a mortality benefit. Because mortality was a secondary objective, our sample was not powered to show significant mortality difference. However, exploratory analysis showed a significant effect on hospital outcome after CDSS use for the neutropenic and severe patients but not for HSCT recipients. Interestingly, SAPSII score was similar at ICU admission between ‘before group’ and ‘after group’ for neutropenic and severe patients. These results should be interpreted cautiously nevertheless they reinforce the hypothesis of a beneficial effect on outcome of using CDSS. After intervention, we described in severe patients, a significant increase in days within protein target, while this effect was not retrieved for calories target. Taken together, these results suggest that protein intake in critically ill hematology patients might be a crucial step of ICU management. Indeed, it has been recently shown in critically ill patients that if the basal amount of protein is provided, restricted calories intake did not impact on outcome [29]. Recent data have also shown in patients who remained in the ICU ≥ 4 days and achieved at least 80% of prescribed protein intake a lower mortality than those achieving < 80% of prescribed. In this situation, percentage of prescribed energy intake was not associated with outcomes [28]. Whether reciprocal relationships exist between calorie and protein intakes exist in severe hematology patients remains to be determined.

Our study presents several limitations. First, the distinction between the acute phase and the recovery period was based on the clinician’s choice. Because there is no consensus on the definitions of acute phase and recovery phase, the retrospective analysis of the clinician’s evaluation might have led to significant variability from one patient to another. Such variability might have influenced the results. Second, due to the retrospective nature of the study, we were unable to collect the causes of non-use of enteral nutrition among the 14 patients who did not benefit. However, the causes of failure of enteral nutrition are detailed Fig. 2. Third, we did not plan to collect cases of refeeding syndrome; however, in our daily practice. these cases are rare, whereas the ICU admission of severely malnourished patients was relatively frequent (22%).

In conclusion, we showed that the use of CDSS allows to increase relatively the rate of days in compliance with caloric and protein targets by more than 50%. Interestingly, compliance rates were low and only reached 30% after intervention. Hospital mortality, ICU-acquired infection and lengths of stay were similar in the two groups.