Introduction
Surgical tracheostomy remains a critical procedure in the elective, critical care and emergency settings.Reference McGrath, Brenner, Warrillow, Pandian, Arora and Cameron1 Indications vary and include airway obstruction, prophylactically securing the airway following major head and neck surgery and in cases of prolonged mechanical ventilation.Reference Cheung and Napolitano2 Surgical complications such as bleeding occur following surgical tracheostomy in approximately 10–27 per cent of cases.Reference Murray, Shen, Massey, Stadler and Zenga3–Reference Ruohoalho, Xin, Bäck, Aro and Tapiovaara5 Additionally these patients may suffer medical complications including lower respiratory tract infections and, rarely, mortality.Reference Murray, Shen, Massey, Stadler and Zenga3–Reference Ruohoalho, Xin, Bäck, Aro and Tapiovaara5
A tracheostomy is distressing for patients and caregivers due to its effect on speech, swallow, appearance, and the associated increased care needs. Although frequently intended as a temporary measure, in some instances a tracheostomy may be permanent.Reference Nakarada-Kordic, Patterson, Wrapson and Reay6 Therefore, understanding an individual patient’s overall health and their anticipated outcome following surgical tracheostomy is vital. This information permits clinicians to better counsel patients and families prior to performing surgical tracheostomy.
To date, authors have looked to describe a host of factors which may portend a worse outcome following surgical tracheostomy. These include technique, timing and patient comorbidity.Reference Brass, Hellmich, Ladra, Ladra and Wrzosek7–Reference Kashlan, Williams, Chang, Yaremchuk and Mayerhoff9 There has been conflicting evidence regarding the effect of chronological age on outcome following tracheostomy.Reference McGrath, Brenner, Warrillow, Pandian, Arora and Cameron10, Reference Levy, Smiley and Latifi11 This likely reflects that chronological age does not accurately reflect a person’s global physiologic state. More recently concepts including frailty have been adopted to distinguish patients’ ability to withstand stressors more accurately than chronological age. Frailty describes individuals who have a reduced physiologic reserve outside the expected process of ageing.Reference Clegg, Young, Iliffe, Rikkert and Rockwood12 While comprehensive geriatric assessment remains the gold standard assessment of frailty, it is time-consuming and resource-intensive, which makes it difficult to implement during the tracheostomy decision-making process.Reference Welsh, Gordon and Gladman13 Subsequently, a host of frailty assessment tools have developed, including the 5-Item Modified Frailty Index.Reference Subramaniam, Aalberg, Soriano and Divino14 This describes frailty according to an accumulation of 5 deficits, congestive heart failure, hypertension, chronic obstructive pulmonary disease and non-independent functional status, making it easy to calculate rapidly and thus readily applicable in clinical practice. Despite the simplicity of the 5-Item Modified Frailty Index, it has been repeatedly demonstrated in other settings to strongly associate with increased surgical morbidity.Reference Wilson, Holzgrefe, Staley, Schenker and Meals15–Reference Tracy, Wilson, Smith, Schenker and Gelbard17
To our knowledge, only one study has assessed morbidity following surgical tracheostomy by frailty status.Reference Chinta, Haleem, Sibala, Kumar, Pendyala and Aftab18 Using the 5-Item Modified Frailty Index and data from the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) the authors demonstrated increased 30-day morbidity following tracheostomy. However, as highlighted by the authors, the findings are limited by use of the ACS-NSQIP database, whichis retrospective, and does not distinguish between percutaneous, surgical or revision tracheostomy. Additionally, the risk of non-decannulation and mortality beyond 30 days could not be assessed using this database.
In this study, we sought to evaluate whether frailty (assessed using the 5-Item Modified Frailty Index) was associated with increased morbidity, failure to achieve decannulation and 6-month mortality after surgical tracheostomy using an institutional database which permitted detailed analysis. Secondarily, we aimed to assess whether any additional patient or surgical factors influenced patient outcomes following surgical tracheostomy.
Methods
Study design and tracheostomy technique
This retrospective cohort study included patients undergoing surgical tracheostomy over a two-year period (2022–2023) at a single institution identified using a prospectively maintained tracheostomy database. Study findings were reported according to the Strengthening The Reporting of OBservational studies in Epidemiology (‘STROBE’) cohort study checklist.Reference von Elm, Altman, Egger, Pocock, Gøtzsche and Vandenbroucke19 Local institutional ethical approval was obtained. Patients eligible for inclusion were aged 18 years or older at the time of tracheostomy. Exclusions were applied to the following groups: (1) percutaneous tracheostomy, (2) emergency cricothyroidotomy, (3) previous tracheostomy, (4) subsequent laryngectomy, and (5) tracheostomy performed prior to index admission.
All surgical tracheostomies in our institution are performed with a horizontal tracheal incision and an upper 2-0 silk stay suture placed at time of insertion. All patients have a cuffed, unfenestrated tube inserted at time of initial surgery which is secured with 2-0 silk sutures and soft tracheostomy neck ties. Following surgical tracheostomy, all patients were reviewed daily by our hospital tracheostomy multidisciplinary care team. Change to non-cuffed and fenestrated tracheostomy is guided by local protocols and clinical need.
Data collection and variables
Data were collected retrospectively from patient healthcare records by two authors (EFC and ANR). Patient demographics and comorbidities were recorded as documented at time of admission. Frailty was recorded as a dichotomous outcome using the 5-Item Modified Frailty Index, with 5-Item Modified Frailty Index scores of 2 or greater equating to frailty. Patient comorbidity was graded using the age-adjusted Charlson Comorbidity Index.Reference Cheung and Napolitano20 The indication for tracheostomy insertion was collected as a dichotomous variable where ‘prolonged ventilation’ included patients undergoing tracheostomy to facilitate ventilatory weaning or in cases of prolonged mechanical ventilation. ‘Airway obstruction’ included patients with acute airway obstruction and elective surgical tracheostomy at time of major head and neck surgery. Cases were classified as elective or emergency.
Binary outcomes recorded included tracheostomy complications such as, bleeding, tracheostomy tube dislodgement, tracheitis and tracheostomy obstructing mucous plug. Complication severity was graded using the Clavien–Dindo classification.Reference Clavien, Barkun, de Oliveira, Vauthey, Dindo and Schulick21 Other outcomes recorded included lower respiratory tract infections, failure to achieve decannulation and six-month post-tracheostomy mortality. Continuous outcomes recorded were time to tracheostomy decannulation and hospital length of stay post tracheostomy tube insertion.
Statistical analysis
Patients were categorised based on 5-Item Modified Frailty Index scores as ‘frail’ (2–5) or ‘non-frail’ (0–1). When analysing patient characteristics, basic descriptive statistics were performed and compared using the chi-square test. Post-tracheostomy outcomes are presented using basic descriptive statistics with differences compared using the chi-square test for dichotomous outcomes. Continuous outcomes (time to decannulation and hospital length of stay) are presented as mean values with corresponding 95 per cent confidence intervals (95 per cent CI) and compared using Student’s t-test. To adjust for potential confounding variables, univariable and multivariable logistic regressions were performed for dichotomous outcomes. Results are displayed as odds ratios (ORs) with corresponding 95 per cent CIs. For continuous outcomes, potential confounding variables were controlled for using univariable and multivariable linear regressions. Results of linear regression analyses are displayed as Beta (β) coefficients with 95 per cent CIs. An alpha level of 0.05 was considered significant. Statistical analysis was performed using STATA BE version 17.0 (StataCorp, College Station, Texas, USA).
Results
Patient characteristics
Of the 201 patients that were identified with a newly formed tracheostomy during the study period, 27 were excluded for the following reasons: percutaneous tracheostomy (n = 18), emergency cricothyroidotomy (n = 1), previous tracheostomy (n = 1), tracheostomy performed with subsequent laryngectomy (n = 7) leaving a total of 174 patients included in our analysis. Twenty-eight (16.1 per cent) patients were classified as frail using the 5-Item Modified Frailty Index. A greater proportion of frail patients were aged above 70 years at time of tracheostomy (p = 0.015). The observed Charlson Comorbidity Index scores were greater among the frail cohort of patients (p < 0.001). The proportion of patients undergoing emergency surgical tracheostomy was greater among frail patients (p = 0.011). A greater proportion non-frail patients had a tracheostomy placed for airway obstruction (p = 0.015) and a greater proportion of non-frail patients were managed in the Intensive Care Unit (ICU) following tracheostomy insertion (p = 0.025). Patient sex, smoking status and anticoagulation usage did not vary by patient frailty (Table 1). The mean hospital length of stay was 42.4 days (95 per cent CI 37.0–49.1 days) and 134 patients were ultimately decannulated (77.0 per cent) with a mean time to decannulation of 21.8 days (95 per cent CI 18.5–25.2 days).
Patient characteristics
5mFI = 5-Item Modified Frailty Index; CCI = Charlson Comorbidity Index; HN = head and neck; ICU = Intensive Care Unit;
* denotes statistical significance
Association of frailty with patient outcomes following surgical tracheostomy
The overall proportion of patients with a tracheostomy-specific complication was 12.1 per cent (21/174) with bleeding the most frequent tracheostomy-related complication (6.3 per cent, 11/174). Most tracheostomy bleeds were managed conservatively (81.8 per cent, 9/11) with two requiring management in theatre. A higher proportion of frail patients had a tracheostomy-related complication (p = 0.008). Frail patients did not suffer more severe tracheostomy-related complications according to the Clavien–Dindo classification (p = 0.854). Specific tracheostomy-related complications (bleeding, tube dislodgement, tracheitis, obstructing mucous plug) did not vary by frail status. Frail patients had a longer time to tracheostomy tube decannulation (p = 0.013) and hospital length of stay (p = 0.004). No difference was observed between groups regarding the incidence of lower respiratory tract infections, the proportion of patients decannulated or six-month mortality (Table 2).
Patient outcomes following tracheostomy stratified by frailty
5mFI = 5-Item Modified Frailty Index; CD = Clavien–Dindo; LRTI = lower respiratory tract infection; 95% CI = 95% confidence interval;
* statistical significance; xchi-square test, unless specified; tStudent’s t-test; asome patients had > 1 tracheostomy-specific complication
Multivariable analysis of complications following surgical tracheostomy
Frailty was the only variable associated with tracheostomy-specific complications on regression analysis (OR 4.09, 95 per cent CI 1.51–11.11; p = 0.006). Frailty was not associated with an increased risk of lower respiratory tract infection. On multivariable analysis, emergent tracheostomy insertion remained independently associated with lower respiratory tract infection (OR 3.08, 95 per cent CI 1.20–7.91; p = 0.019), while patients with a tracheostomy inserted for airway obstruction were less likely to develop a lower respiratory tract infection (OR 0.32, 95 per cent CI 0.15–0.68; p = 0.003). Frailty was not associated with an increase in tracheostomy decannulation failure or post-tracheostomy six-month mortality. Multivariable analysis demonstrated that tracheostomies inserted for airway obstruction (OR 4.62, 95 per cent CI 1.95–10.95; p = 0.001) and those patients managed in ICU post-operatively (OR 5.62, 95 per cent CI 2.17–14.57; p < 0.001) were more likely to be decannulated successfully. On multivariable logistic regression, increased Charlson Comorbidity Index scores were associated with increased six-month post-tracheostomy mortality (OR 1.96, 95 per cent CI 1.02–3.77; p = 0.043). Patients undergoing surgical tracheostomy to bypass upper airway obstruction had lower six-month mortality following tracheostomy compared to those undergoing tracheostomy for prolonged ventilation (OR 0.25, 95 per cent CI 0.09–0.66; p = 0.005). Table 3 demonstrates the detailed univariable and multivariable logistic regression analyses.
Univariable and multivariable logistic regression analysis of complications following tracheostomy
OR = odds ratio; 95% CI = 95% confidence interval; CCI = Charlson Comorbidity Index; AM = airway management; ICU = Intensive Care Unit; LRTI = lower respiratory tract infection;
* statistical significance
On multivariable linear regression, frailty was not independently associated with time until tracheostomy decannulation (β 5.29 days, 95 per cent CI −2.70–13.27 days; p = 0.192). Patients with a tracheostomy inserted to bypass upper airway obstruction had a shorter time to tracheostomy decannulation on multivariable linear regression compared to those undergoing tracheostomy for prolonged ventilation (β −26.61 days, 95 per cent CI −34.47 to −18.74; p < 0.001). Frailty was independently associated with increased hospital length of stay (β 15.76 days, 95 per cent CI 1.06–30.44 days; p = 0.036). Patients with a tracheostomy inserted to bypass upper airway obstruction had decreased hospital length of stay on multivariable linear regression compared to patients undergoing tracheostomy for prolonged ventilation (β −23.61 days, 95 per cent CI −37.35 to −9.87; p = 0.001). Table 4 displays univariable and multivariable linear regression analyses.
Univariable and multivariable linear regression analysis of time to decannulation and hospital length of stay following tracheostomy
95% CI = 95% Confidence Interval; CCI = Charlson Comorbidity Index; AM = Airway Management; ICU = Intensive Care Unit;
* statistical significance
Discussion
After adjusting for potential confounding variables such as patient age or comorbidity, the present study demonstrates that frailty, as described using the 5-Item Modified Frailty Index, is independently associated with a 4× increased risk of tracheostomy-specific complications and a 15-day increased hospital length of stay following surgical tracheostomy. These findings support the only other study known to us of frail patients undergoing tracheostomy but use detailed prospectively maintained individual patient data rather than a national database allowing analysis of a more homogenous cohort including only surgical tracheostomy with longer follow up data.Reference Chinta, Haleem, Sibala, Kumar, Pendyala and Aftab18 Intuitively, the present study’s findings are unsurprising, given the presence of frailty indicates a person’s decreased ability to respond to stressors, such as tracheostomy insertion.Reference Walston and Bandeen-Roche22 Indeed, the 5-Item Modified Frailty Index has demonstrated a strong association with increasing post-operative morbidity after head and neck surgery.Reference Cleere, Davey and O’Neill23, Reference Goshtasbi, Birkenbeuel, Lehrich, Abiri, Haidar and Tjoa24 The findings of the present study would suggest that frailty, as defined by the 5-Item Modified Frailty Index, may be a useful marker to stratify a subgroup of patients who can be expected to have a more challenging and prolonged hospital post-operative course following surgical tracheostomy. Thus 5-Item Modified Frailty Index scores may guide discussions among clinicians, patients and their families prior to surgical tracheostomy.
The development of lower respiratory tract infections, six-month mortality, tracheostomy decannulation and time to tracheostomy decannulation was not significantly affected by the presence of frailty. Conversely, patients with a surgical tracheostomy inserted for airway obstruction had lower rates of lower respiratory tract infection, six-month mortality and were more likely to be decannulated successfully relative to patients undergoing tracheostomy for prolonged ventilation. Additionally, patients receiving a surgical tracheostomy for airway obstruction had reduced time to tracheostomy tube decannulation and hospital length of stay. Tracheostomy-related surgical complications were not altered by the indication for tracheostomy insertion.
While not the primary aim of the present analysis, these findings are noteworthy. Patients undergoing tracheostomy for prolonged ventilation are typically critically ill ICU patients with poor expected health-related outcomes.Reference Andriolo, Andriolo, Saconato, Atallah and Valente25, Reference Mallick and Bodenham26 Patients undergoing surgical tracheostomy for prolonged ventilation were more likely to be frail in the current study. As such, the present study’s findings may indicate that the prolonged ventilation patient group was a more unwell, frail cohort rather than specifically the effect of tracheostomy indication on post-surgical tracheostomy morbidity. Clear, open discussions with with family members of sedated and ventilated ICU patients proposed for tracheostomy are essential prior to performing the surgical procedure. Our findings may aid clinicians in these discussions and allow clinicians to highlight the increased post-tracheostomy morbidity in this patient population.
Nearly one-in-six patients in the present study were classified as frail using the 5-Item Modified Frailty Index. The proportion of frail patients was notably higher with increasing age (11/42, 26.2 per cent > 70 years vs 2/45, 4.4 per cent < 55 years; p = 0.015) which is consistent with available literature.Reference Fogg, Fraser, Roderick, de Lusignan, Clegg and Brailsford27, Reference Piccirillo, Vlahiotis, Barrett, Flood, Spitznagel and Steyerberg28 Unsurprisingly, frail patients were also more comorbid relative to their non-frail counterparts using the Charlson Comorbidity Index (p < 0.001), which again corroborates existing literature.Reference Wong, Weiss, Sourial, Karunananthan, Quail and Wolfson29
The present study demonstrates that frail patients were more likely to undergo emergency surgical tracheostomy. This may be due to frail patients decreased physiologic reserve and potentially frail patients resultant decreased tolerability of a narrowed upper airway, thus, necessitating emergent tracheostomy on patient presentation.Reference Murray, Shen, Massey, Stadler and Zenga30 However, this likely does not solely explain this finding as some patients may be managed with initial emergent intubation, without conversion to tracheostomy to stabilise the airway.Reference Law, Duggan, Asselin, Baker, Crosby and Downey31 An additional potential rationale for the increase in emergency tracheostomies may be due to clinician decision making being altered in cases involving perceived frail patients. Previous literature has demonstrated that frailty influences clinician decision making across a variety of medical and surgical disciplines.Reference Brar, Whitlock, Komenda, Rigatto, Prasad and Bohm32–Reference Khatry, Peel, Gray and Hubbard35 Unfortunately, owing to the retrospective nature of the present study, the rationale for individual clinician decision making on a case-by-case basis was not possible to assess. Future work would benefit from exploring the reasons for increased emergency tracheostomies in frail patients and clinician attitudes surrounding tracheostomy insertion among frail and non-frail patients.
The present study does have several limitations. The retrospective nature of the study design may introduce uncertainty and confounding bias. Univariable and multivariable regression analyses were performed to address this. Secondly, we excluded patients undergoing percutaneous and non-primary tracheostomy. Therefore our findings should not be used to draw conclusions on the effect of frailty in these populations. The present study was unable to assess the effect of surgical tracheostomy on speech and swallow outcomes. Future evaluations of frail patients undergoing surgical tracheostomy may benefit from these analyses. Finally, the number of frail patients (n = 28) in the present study is small and the overall rate of surgical tracheostomy complications (12.1 per cent, 21/174) is low. As such, analysis of specific tracheostomy-related complications was underpowered.
In conclusion, frailty assessed using the 5-Item Modified Frailty Index may be used to detect patients more likely to suffer tracheostomy-specific morbidity and those potentially requiring a longer hospital length of stay after surgical tracheostomy. A tracheostomy indication to manage airway obstruction is strongly associated with reduced risk of lower respiratory tract infections, permanent tracheostomy, six-month mortality and reduced time to decannulation and hospital length of stay. By recognising patient frailty and tracheostomy indication prior to surgical tracheostomy, clinicians may better appreciate expected outcomes which may guide their decision making.
• The effect of frailty on surgical tracheostomy is under reported
• In the present study, approximately one-in-six (16.1 per cent) patients undergoing surgical tracheostomy were frail
• Overall, 12.1 per cent of patients suffered complications directly related to surgical tracheostomy placement
• Frail status was independently associated with longer hospital length of stay and tracheostomy-specific complications post-operatively
• Frail status was not associated with an increased likelihood of failed tracheostomy decannulation, time to tracheostomy decannulation or six-month mortality after surgical tracheostomy
Data availability
Data or code available on reasonable request from authors.
Funding
No funding was obtained for the purposes of this study.
Conflicts of interest
The authors have no conflicts of interest to declare.
Ethics/Consent
This study was approved by the St James’s Research and Innovation Office who waived the requirement for informed consent as this was a retrospective evaluation of anonymised data.
Author contributions
Conceptualisation: EF Cleere, CVI Timon, J Kinsella, P Lennon, CWR Fitzgerald. Study design: EF Cleere, JM Hintze, CWR Fitzgerald. Data acquisition: EF Cleere, AN Ramesh, I O’Riordan. Data analysis and interpretation: EF Cleere, JM Hintze, CWR Fitzgerald. Statistical analysis: EF Cleere, JM Hintze. Writing original draft: EF Cleere. Writing, review and editing: all authors. Supervision: CWR Fitzgerald.
