Tracheotomy does not affect reducing sedation requirements of patients in intensive care – a retrospective study

Intubated and mechanically ventilated patients in the intensive care unit (ICU) often need sedation to treat anxiety, agitation and/or pain. Indeed, prolonged translaryngeal intubation is a major source of physical discomfort, for which sedation often is necessary [1‐5]. In addition, sedation may be required to facilitate mechanical ventilation (MV). However, sedation leads to a depression of cardiovascular, respiratory and immunological systems, and prolongs the time spent on the ventilator; it can therefore influence outcome in critically ill patients [6‐11].

Tracheotomy is a common procedure in mechanically ventilated patients in the ICU, especially in those who have an (expected) prolonged duration of MV [12‐15]. Tracheotomy is frequently performed 7 to 21 days after the start of MV [14]. A recent meta-analysis suggests that early tracheotomy shortens the duration of artificial ventilation and length of stay in the ICU [16]. Current opinion is that tracheotomy reduces sedation requirements by facilitating oral and bronchopulmonary toilet and increasing patient comfort [1, 17, 18]. One recent study described the influence of tracheotomy on the amount of sedation needed in critically ill patients [19]. However, the influence of timing of tracheotomy and differences between subgroups have never been examined.

The objective of this single-center observational study was to determine sedation requirements (amount of sedation used and number of mechanically ventilated patients requiring sedation) in our mixed medical–surgical ICU before and after tracheotomy. In addition, we investigated whether this was influenced by the timing of tracheotomy and patient category, namely acute surgical, elective surgical and non-surgical patients.

Of 1,788 patients admitted during the study period, 129 (7%) received a tracheotomy. Patients who received a tracheotomy on the day of admittance and patients who received a tracheotomy in another hospital before they were transmitted to our ICU were excluded from analysis, as were those who were not intubated before they received tracheotomy. This left us with 117 patients for the present analysis. Patients received a tracheotomy after a median time of nine (interquartile range 5 to 14) days from the start of MV. Patient characteristics are shown in Table 1. Basic information about the amount of sedatives used in our population is shown in Table 2.

It has been suggested that tracheotomy reduces the sedation requirements of ICU patients. This retrospective observational study shows, however, that sedation requirements were already sharply declining before tracheotomy, and that tracheotomy did not further reduce sedation needs. Similarly, the number of patients requiring sedation also decreased gradually before tracheotomy, and there was no significant difference in the number of patients receiving morphine, midazolam or propofol during the two days before tracheotomy compared with the two days after tracheotomy.

There are several limitations to our study. First, it was a retrospective analysis. Randomized trials are needed, to compare groups of patients. It is, however, very difficult to match a control group that is exactly the same as the patient group except for not receiving tracheotomy. Second, we did not collect data on sedation scores of each patient. Third, as in the study of Nieszkowska and colleagues [19], we did not implement in our sedation protocol a systematic withdrawal of sedation. This could have influenced the need for sedation before and after tracheotomy [7]. Finally, as mentioned above, this study was performed in a single-center ICU, which makes it difficult to generalize data to other ICUs.

Recent studies have shown that the use of excessive sedation prolongs the duration of MV and the length of stay in ICU, and negatively affects patient outcome [7‐11, 22]. It is therefore important to use strict guidelines that aim at the reduction of sedation whenever possible. Compliance with protocols in our unit is high because there is good communication between doctors and nurses. Every day the policy for each patient is discussed and nurses therefore do not have much scope for deviation from the protocols. Non-compliance with sedation protocols usually leads to oversedation [7]. This might influence the effect of tracheotomy on sedation use. A recent meta-analysis suggests that early tracheotomy shortens the duration of artificial ventilation and length of stay in the ICU [16]. In addition, Rumbak and colleagues [23] showed that early tracheotomy, as compared with prolonged endotracheal intubation, shortens the duration of MV in ICU patients and decreases the number of days spent sedated.

Nieszkowska and colleagues [19] recently described in more detail the influence of tracheotomy on the amount of sedation needed in critically ill patients. In contrast to our findings, they showed a significant decrease in sedation requirements of fentanyl and midazolam after tracheotomy. There are several possible explanations for the discrepancy between our study and that of Nieszkowska and colleagues. The first might be differences in the applied sedation protocols. In our institution the adequacy of sedation is determined by the SEDIC score. This scale was developed in 2003 by our department by combining elements of existing sedation scales, including the Riker sedation agitation scale (SAS) used in the study of Nieszkowka and colleagues. At present, no studies have been published that compare these two sedation protocols directly. The reliability of the SEDIC scale is supported by the inter-observer agreement of simultaneous but independently assessed SEDIC scores between the attending nurse and a research nurse (intra-class correlation coefficient 0.88; 95% confidence interval 0.81 to 0.92). The validity was assessed first by the observed agreement between score categories (level of stimulus and response) with a weighted kappa of 0.82, second by the ability of the SEDIC score to predict the time needed to wake up after terminating the sedative (r² 42%), and third by the association between the Ramsay scale and the SEDIC scale, Spearman rank-order correlation coefficient of 0.74 (p = 0.1) [20]. In the study of Nieszkowska and colleagues, adequacy of sedation is determined every three to four hours, in contrast with every two hours in our study. It is therefore possible that with the SEDIC score deep sedation levels are noticed and corrected for at an earlier stage. In addition, the SEDIC score might be more sensitive to the detection of inappropriately high sedation in ICU patients [20].

Importantly, Nieszkowska and colleagues did not correct for individual differences in need for sedation between patients. We explicitly divided the daily dose by the mean daily dose of all days measured for each patient (DD/MDD). In this way we could overcome the possibility that some patients carried relatively more weight in the study and hence overestimated the need for sedation, especially before tracheotomy. Our timing of tracheotomy was much earlier (median nine days) than that in the study of Nieszkowska and colleagues (median 14 days). Accordingly, the timing of tracheotomy in our study cannot explain the discrepancy between the two studies. Finally, there might be an important difference in case-mix between the two studies, causing a difference in the amount of sedation needed and thereby a difference in the impact of tracheotomy on sedation. In the former study, 40% of the patients were admitted for shock and the number of cardiac patients was much higher in their study (52.7%) than in ours (13.7%). In addition, there was an important difference in ICU mortality: 30.5% versus 12.8% in our study. Accordingly, it is very difficult to generalize the results of these two single-center ICU studies because of the possible differences in sedation practice and in MV and weaning protocols between the different ICU settings.

We found that patients in the non-surgical and acute surgical groups had higher dosages of midazolam than patients in the elective surgical group, but there was no difference in the rate of decline before or after tracheotomy between patient groups. Importantly, we found no influence of timing of tracheotomy on sedation requirements. Consequently, the high decline in sedation before tracheotomy cannot be explained by the timing of tracheotomy. In both early and late tracheotomy patients the dosage remained stable after tracheotomy.

We found no convincing support for the hypothesis that tracheotomy reduced the sedation needs of critically ill patients in our ICU. In our center, using a strict sedation protocol, sedation requirements were already sharply declining before tracheotomy was performed.