Suprastomal tracheal stenosis after dilational and surgical tracheostomy in critically ill patients


Assen Koitschev, MD


We have previously reported cases of severe suprastomal stenosis after tracheostomy. In this observational study we investigated the occurrence of suprastomal stenosis as a late complication. Patients with persistent tracheostomy after intensive care underwent an endoscopic examination of tracheostoma, larynx and trachea. A percutaneous dilational tracheostomy was employed in 105 (71.9%) and surgical tracheostomy in 41 (28.1%) of the cases (n = 146). The incidence of severe suprastomal stenosis (grade II > 50% of the lumen) was 23.8% (25 of 105) after dilational tracheostomy and 7.3% (3 of 41) after surgical tracheostomy (p = 0.033). Age, gender, underlying disease, ventilation time, and swallowing ability were not significantly associated with the tracheal pathology. This study suggests that dilational tracheostomy is associated with an increased risk of severe suprastomal tracheal stenosis compared to the surgical technique.

The indications for a tracheostomy include anticipated prolonged respiratory assistance, access for management of lower airway secretions and the prevention of injury to the larynx [1]. Within the last two decades, open surgical tracheostomy has been increasingly replaced by percutaneous dilational tracheostomy [2]. Reports suggest that dilational tracheostomy is a more efficient and cost-effective technique [2–4]. Dilational techniques have also allowed intensive care physicians to perform tracheostomies at the bedside without depending on a surgical team [5–7]. The introduction of flexible bronchoscopy has made tracheostomy easy to teach with minor intra-operative complications in the hands of an experienced team [8, 9]. However, much less is known about the potential late complications of dilational tracheostomy compared with the surgical technique [4, 10].

The commonest arguments in favour of dilational tracheostomy are reduced operative time, ease of performance, ability to be performed at the bedside and lower cost [11, 12]. However, if flexible bronchoscopy is performed, both procedures require sedation, special equipment and assistance, bringing into question some of the advantages proposed for dilational tracheostomy [13]. In fact, all tracheostomy techniques can be performed at the bedside with relative ease [14]. Nevertheless, the perceived advantages of dilational techniques make this procedure the first choice [15] for the critically ill patient needing long-term airway control [16].

Whereas most tracheostomies are required for short-term airway control, or the patients do not survive their acute hospital admission, an increasing number of patients with a tracheostoma are discharged from the intensive care unit prior to decannulation. Few intensive care services carry out long-term follow-up of these patients [15]. Rehabilitation specialists may be responsible for them, which may account for the relative paucity of reports in the critical care literature dealing with a long-term follow-up of patients with a persisting tracheostoma [17]. Recent case reports, including our own data, suggest that there may be a high incidence of suprastomal tracheal stenosis as a late complication after dilational tracheostomy [18, 19]. The goal of this study was to investigate the risk of developing a suprastomal stenosis in patients with a persisting tracheostoma.


Patients and data

An observational study was performed in three facilities specialising in postintensive care rehabilitation. Patients were referred from 52 specialised or multidisciplinary intensive care units over a large area in Eastern Germany. The only criterion for a transfer was the need for further rehabilitation after the acute phase of treatment. All patients were treated initially in large referral centres with intensive care facilities, where the tracheostomy procedures were performed. Once the patients did not require any intensive care management and were self ventilating, they were transferred. Most patients required a permanent tracheostomy due to their underlying neurological impairment. Over a 32-month period, all patients with a persistent tracheostoma admitted following intensive care underwent endoscopic examination of their tracheostoma, larynx and trachea. In all patients an assessment of the tracheostomy canal and the trachea above and below the stoma was possible after removal of the cannula. The examination was done uniformly and reproducibly using 4-mm angled rigid and/or flexible endoscopes, which allowed examination of the trachea even in cases of a stenosis. The grading of the stenosis was made by an experienced laryngologist based on a two-dimensional estimation of the cross-sectional area of the persisting suprastomal lumen compared with the supposed diameter of the normal trachea.

The study was approved by the Research Ethics Committee of the University of Rostock. All patients and/or the responsible relatives gave their consent and data was anonymised before statistical evaluation. The doctors performing the endoscopic evaluation were blinded to the method of tracheostomy.

Swallowing was evaluated according to the patient's subjective judgement and was considered normal if a feeding tube was not needed to maintain adequate dietary intake. Voice was examined by introducing a speech cannula or by temporarily blocking the tracheostomy. After the endoscopy, the referring facilities were contacted with a questionnaire about the patient's diagnosis, the indication for the tracheostomy, the date, and the tracheostomy procedure employed. Of the 169 patients examined, only the data of the 146 patients for whom the tracheostomy technique was known were analysed.

The following data were recorded: age, gender, pathological findings and their localisation (larynx, trachea and tracheostoma), vocalisation and swallowing ability. The suprastomal tracheal stenosis, as primary outcome, was graded according to Hoppe et al. [20]: grade I if less than 50% and grade II if more than 50% of the lumen was obstructed.

Definition of tracheostomy techniques

We defined two principle types of tracheostomy: dilational tracheostomy and open surgical tracheostomy. Dilational techniques are known by their inventor's publications [5–7] and referred to as percutaneous dilational tracheotomies. All these procedures have the common principle of a flexible guidewire being inserted into the trachea via a needle under bronchoscopic monitoring. Single or multiple dilators of increasing diameter are then passed along this wire, generating a tracheal opening. Usually the tracheostomy tube is introduced over the last dilator passed.

As a common principle of elective open surgical tracheostomy, the trachea is exposed via a skin incision above the jugulum. However, the incision technique used for opening the anterior tracheal wall may vary. In the classical form of a tracheostomy a horizontal incision between the second and third tracheal ring is used. A horizontal H-incision based on second and third tracheal ring, a simple vertical incision or an anterior resection of a single tracheal ring are variations of the classical tracheostomy and are summarised and referred to as non-flap tracheostomy techniques [21]. In contrast to the non-flap surgical tracheostomy, the incision used for the anterior tracheal wall in a flap-tracheostomy consists of an inferior-based flap derived from the anterior part of the second or third tracheal ring. The so-called inverted U-flap or Bjork flap [22] is then sutured to the inferior skin margin of the horizontal skin incision. This technique creates a mucocutaneous junction and results in a stable and epithelialised tracheostoma. This technique is also referred to as flap-tracheostomy.


Median, interquartile range (IQR) and mean values (SD) were calculated. The characteristics of the two groups were compared using the two-tailed Fisher's exact test, two-sample t-test or Mann–Whitney U-test as appropriate. To compare the time between tracheostomy and endoscopic examination, the non-parametric Kruskal–Wallis test was employed. The level of statistical significance was set at p < 0.05. Unifactorial analysis was performed for all documented factors (age, gender, underlying disease, ventilation time before tracheostomy, period between tracheostomy and examination, phonation ability and swallowing ability). Variables that reached potential statistical relevance (p < 0.10) were included in further multifactorial modelling by nominal logistic regression analysis. All statistical analyses were carried out with SPSS® software, release 12.0.1 (SPSS Inc., Chicago, IL).


The data for 146 patients were evaluated. In all cases the indication for tracheostomy was anticipated prolonged ventilation during intensive care. Information about the characteristics of the patients, general disease, indication, technique, and time of tracheostomy are summarised in Table 1.

Table 1.   Main characteristics of the study population by tracheostomy technique.
 Type of tracheostomypAll
  • *

    Mann–Whitney U-test was taken due to non-normal distribution of the values. IQR, interquartile range; ST, surgical tracheostomy; PDT, percutaneous dilational tracheostomy.

n10541 146
Age; mean (SD)57.4 (15.3)58.6 (13.6)0.68057.7 (14.8)
Sex; n (%)
 F38 (36.2%)14 (34.1%)0.85052 (35.6%)
 M67 (63.8%)27 (65.9%) 94 (64.4%)
Diagnosis; n (%)
 Neurological disease65 (61.9%)25 (61.0%)0.97190 (61.6%)
 Internal disease20 (19.0%)7 (17.1%) 27 (18.5%)
 Trauma20 (19.0%)7 (17.1%) 27 (18.5%)
 Others0 (0.0%) 2 (4.9%) 2 (1.4%)
Ventilation prior to tracheostomy
[days]; median (IQR)
6 (4–10)8 (3–14)0.426*6.5 (4–10)
Time between tracheostomy
and exam [days]; median (IQR)
52 (28–100)55 (31–76)0.976*52.5 (28–92)

A dilational technique was used in 105 patients (71.9%) and a surgical technique in 41 (28.1%). Assuming all postintensive care patients have an equal chance of entering rehabilitation facilities prior to decannulation, the data indicate a preference to use dilational methods to perform a tracheostomy across the referral base of 52 varied intensive care facilities.

In 131 (89.7%) patients the duration between the tracheostomy and the endoscopic examination was documented and ranged from 1 to 587 days (mean, 75.0; SD, 82.6; median, 52.5).

In 95 (65.1%) patients the duration of assisted ventilation before tracheostomy could be retrieved and ranged from 1 to 41 days (mean, 8.4; SD, 7.0; median, 6.5). The frequency of dysphagia, as an indirect sign of possible chronic aspiration, was 52.5%. The majority of the patients (60.3%) were able to vocalise.

The outcome of interest was the presence or absence of stenosis (grades I and II). It is important to note that the tracheal stenosis was suprastomal in all cases. A stenosis was found in 48.6% of the patients of which severe stenosis (grade II) was seen in 28 (19.2%). Phonation disability and dysphagia were more frequent in the stenosis group but the difference did not reach significance (p = 0.128 and p = 0.130, respectively). Only two variables appeared to be potentially relevant – the technique of tracheostomy (p = 0.0004) and postoperative time (p = 0.040).

Tracheostomy technique

The occurrence of suprastomal stenosis by surgical technique is shown in Table 2. Stenosis was twice as frequent after a dilational tracheostomy compared to surgical tracheostomy (58.1%vs 24.4%; p = 0.0004). The risk of developing a severe tracheal stenosis (more than 50% of the tracheal lumen) was found to be significantly higher with the dilational tracheostomy than the surgical tracheostomy (23.8%vs 7.3%; p = 0.033).

Table 2.   Occurrence of suprastomal stenosis by surgical technique.
 Type of tracheostomypAll
  1. ST, surgical tracheostomy; PDT, percutaneous dilational tracheostomy.

n10541 146
Suprastomal lumen; n (%)
 Normal44 (41.9%)31 (75.6%)0.000475 (51.4%)
 Stenosis Grade I 36 (34.3%) 7 (17.1%)  43 (29.5%)
 Stenosis Grade II 25 (23.8%) 3 (7.3%)  28 (19.2%)

The dilational and surgical group were compared with respect to underlying disease, gender and age, ventilation time prior to the surgical procedure, and time between tracheostomy and endoscopic observation. No significant differences were found between any of the categories (Table 1). The frequency of dysphagia, as an indirect sign of possible chronic aspiration, was similar in both groups (dilational tracheostomy 54.5%, surgical tracheostomy 47.5%). In contrast, the ability to phonate was more frequently present in the surgical tracheostomy group (dilational tracheostomy 54.3%, surgical tracheostomy 75.6%, p = 0.024).

Postoperative time

The time between the tracheostomy and endoscopic observation was significantly longer when a suprastomal stenosis was present (p = 0.001, grade II only; p = 0.040 grade I and grade II). This has been further analysed with respect to both tracheostomy technique and grade of suprastomal stenosis (Table 3). There was a difference in postoperative time between grade of stenosis (grade I, grade II, normal) groups (Kruskal–Wallis test, p = 0.004). However, no significant difference could be seen after surgical tracheostomy (p = 0.422). In contrast, postoperative time did differ significantly between grade of stenosis groups after dilational tracheostomy (p = 0.009). Most notably, after dilational tracheostomy there was a marked time difference between the grade I group (median 46.5 days) and the grade II group (median 92 days) (p = 0.014).

Table 3.   Postoperative time (days) by technique of tracheostomy and grade of suprastomal stenosis.
 Time between tracheostomy
and exam [days]; median (IQR)
  1. IQR, interquartile range; ST, surgical tracheostomy; PDT, percutaneous dilational tracheostomy.

Normal/no stenosis39 (21–70)52 (30–75)
Stenosis grade I46.5 (26–92)60.5 (26–91)
Stenosis grade II92 (54–167)69 (59–143)

The risk of finding a suprastomal stenosis was also examined with logistic regression analysis. Modelling both the overall risk of stenosis (grade I and grade II) and the risk of severe stenosis (grade II), only the technique of tracheostomy and postoperative time period remained in the models, the surgical technique (dilational vs open surgical) being an independent predictor of suprastomal tracheal stenosis.


In this study we present data on critically ill patients who received tracheostomies for long-term ventilation. The group of patients studied is of particular interest as all patients survived and were successfully weaned from mechanical ventilation but still needed a tracheostomy. We have previously demonstrated that a persisting tracheostoma after percutaneous dilational tracheostomy may lead to severe suprastomal stenosis and/or obliteration of the trachea [18]. As a consequence, we conducted this observational study to evaluate the incidence of this detrimental complication in a larger cohort of patients.

In 1985, Ciaglia re-introduced the technique of percutaneous dilational tracheostomy, which consists of the insertion of a guidewire through a puncture of the trachea, followed by several ‘dilators’ to create a sufficiently wide stoma [5]. Since then, the use of different variations of non-surgical dilational tracheostomies has dramatically increased [7, 9, 23–26]. At least four commercial sets for such ‘minimal invasive’ procedures are currently in use. In three of them, the cannula is inserted from outside. The other technique described by Fantoni and Ripamonti [6] in 1997 uses a one-step procedure combining dilation and cannula insertion. The device is introduced through the larynx and has to be rotated in the trachea to reach the correct position.

Percutaneous dilational tracheostomy is a procedure widely accepted by intensive care physicians. Studies comparing the versatility of open surgical vs dilational tracheostomy techniques provide compelling evidence that the latter saves time and expense and is easy to perform at the bed side [27]. However, little is known about the potential long-term complications after percutaneous dilational tracheostomy, especially in patients with a persisting tracheostoma.

Tracheal stenosis after tracheostomy was observed cranial to the tracheostoma and consisted of scar and granulation tissue. We speculate that the curved shape of the dilators and the recommended downward orientation of the dilator tip in the trachea used to avoid injury to the posterior tracheal wall results in an inward tearing of tracheal cartilage and soft tissues above the dilation site and an outward tearing below the dilation site. The outward pressure on the caudal anterior wall and an inward pressure on the cranial anterior wall of the trachea persist by the tracheal cannula due to the angle and the caudal direction of the tracheostomy canal. Therefore the intrusion of the anterior wall might be fixed in this position by the pressure of the inserted cannula. This could explain the frequently observed suprastomal tracheal stenosis.

In some of the classical non-flap surgical techniques such as simple horizontal incision of the anterior tracheal wall, a similar inward tearing mechanism of the suprastomal anterior tracheal wall can be assumed. This classical technique was used in all three cases of grade II stenosis seen in our patients after surgical procedures. In fact, stenosis as a complication associated with tracheostomy was described at the end of the 19th and the beginning of the 20th century before the introduction of translaryngeal intubation [28–30]. In those cases the tracheostoma was usually created by a classical open tracheostomy technique without a flap or by the use of ‘tracheotom’ instruments which appear very similar to the dilators used in the percutaneous dilational technique today.

In contrast to dilational tracheostomy and non-flap surgical tracheostomy, a flap tracheostomy with an inverted U-flap of the anterior tracheal wall seems to minimise the risk for the development of a severe suprastomal tracheal stenosis. The creation of a stable and epithelialised tracheostoma may help to avoid inward tearing of the suprastomal anterior tracheal wall, thereby decreasing the risk for the development of a suprastomal tracheal stenosis as a long-term complication.

Our data are derived from a comparison of two groups of multimorbid patients who survived a variety of serious diseases treated in a variety of intensive care units. There are likely to be a multitude of individual factors contributing to patient outcome that have not been included in this retrospective evaluation. We could speculate that the experience of the surgeons and other health care personnel, as well the appropriateness of the tracheal toilet and tubes, may influence the development of stenosis.

The time between tracheostomy and endoscopic observation was significantly longer in the stenosis group, and differed significantly between the grades of stenosis groups (grade I vs grade II). The risk of severe stenosis (grade II) is clearly increased with the duration of the tracheostomy but the postoperative time difference between grade I and grade II stenosis was only significant in the dilational tracheostomy group. According to our clinical experience and data published by others, dyspnoea could be anticipated in all cases with a stenosis of grade II and higher [31, 32]. Early decannulation may have contributed to the excellent outcome of dilational techniques described in other studies.

Based on the results of this study it appears that an increased risk for developing a suprastomal tracheal stenosis as a late complication of dilational tracheostomy is a consequence of a combination of this particular technique and the persistence of a tracheostoma. Stenosis of more than 10% of the cross-sectional area of the trachea as a long-term complication in decannulated patients was found in 26% of the 54 patients examined by computer tomography. Only one patient revealed a stenosis of more than 50%. The site of stenosis was 25–70 mm below the vocal cords, corresponding to the stoma and suprastomal area [33].

Severe tracheal stenosis complicates or even prevents successful rehabilitation. Multiple revision surgeries were necessary in many of our patients, including converting dilational tracheostoma into a flap tracheostoma and/or removal of granulations. In one case a tracheal resection with an end-to-end tracheal anastomosis was necessary. In a second case, major comorbidities prevented the patient from undergoing surgical management of complete suprastomal obliteration of the trachea. This patient had to be provided with an electronic voice device.

The decision about which tracheostomy to perform should include an estimate of the potential time for which an individual patient might require a tracheostoma. If the severity of the underlying disease implies a longer period without decannulation, a flap-tracheostomy should be performed. We believe that this would help to minimise the risk of complications, optimise tracheostomy care and improve the ability of the patient to communicate during rehabilitation. Our personal experience suggests that the creation of a stable and epithelialised tracheostoma may help to avoid inward tearing of the suprastomal anterior tracheal wall, thereby decreasing the risk of developing a suprastomal tracheal stenosis. In particular this may be relevant when neurological deficits such as dysphagia and/or chronic aspiration are anticipated.

Endoscopic examination in patients with a tracheostomy may also be useful, particularly when accidental decannulation occurs, as an emergency airway situation may occur if the possibility of a suprastomal tracheal stenosis has not been excluded.


The authors thank Mrs D. Guénon (Department of Medical Information Processing, University of Tuebingen) for the review of the statistical data and the help with data analysing. The study received no external financial support.