BackgroundThe COVID-19 pandemic has strained intensive care unit (ICU) resources. Tracheotomy is the most frequent surgery performed on ICU patients and can affect the duration of ICU care. We studied the association between when tracheotomy occurs and weaning from mechanical ventilation, mortality, and intraoperative and postoperative complications.MethodsMulticentre prospective cohort including all COVID-19 patients admitted to ICUs in 36 hospitals in Spain who received invasive mechanical ventilation and tracheotomy between 11 March and 20 July 2020. We used a target emulation trial framework to study the causal effects of early (7 to 10 days post-intubation) versus late (>10 days) tracheotomy on time from tracheotomy to weaning, postoperative mortality, and tracheotomy complications. Cause-specific Cox models were used for the first two outcomes and Poisson regression for the third, all adjusted for potential confounders.FindingsWe included 696 patients, of whom 142 (20·4%) received early tracheotomy. Using late tracheotomy as the reference group, multivariable cause-specific analysis showed that early tracheotomy was associated with faster post-tracheotomy weaning (fully adjusted hazard ratio (HR) [95% confidence interval (CI)]: 1·31 [1·02 to 1·81]) without differences in mortality (fully adjusted HR [95% CI]: 0·91 [0·56 to 1·47]) or intraoperative or postoperative complications (adjusted rate ratio [95% CI]: 0·21 [0·03 to 1·57] and 1·49 [0·99 to 2·24], respectively).InterpretationEarly tracheotomy reduced post-tracheotomy weaning time, resulting in fewer mechanical ventilation days and shorter ICU stays, without changing complication or mortality rates. These results support early tracheotomy for COVID-19 patients when clinically indicated.FundingSupported by the NIHR, FAME, and MRC.Research in contextEvidence before this studyThe optimal timing of tracheotomy for critically ill COVID-19 patients remains controversial. Existing guidelines and recommendations are based on limited experiences with SARS-CoV-1 and expert opinions derived from situations that differ from a pandemic outbreak. Most of the available guidance recommends late tracheotomy (>14 days), mainly due to the potential risk of infection for the surgical team and the high patient mortality rate observed early in the first wave of the COVID-19 pandemic.Recent publications have shown that surgical teams can safely perform tracheotomies for COVID-19 patients if they use adequate personal protective equipment. Early tracheotomy seems to reduce the length of invasive mechanical ventilation without increasing complications, which may release crucial intensive care unit (ICU) beds sooner.The current recommendations do not suggest an optimal time for tracheotomy for COVID-19 patients, and no study has provided conclusions based on objective clinical parameters.Added value of this studyThis is the first study aiming to establish the optimal timing for tracheotomy for critically ill COVID-19 patients requiring invasive mechanical ventilation (IMV). The study prospectively recruited a large multicentre cohort of 696 patients under IMV due to COVID-19 and collected data about the severity of respiratory failure, clinical and ventilatory parameters, and whether patients need to be laid flat during their ICU stay (proned). The analysis focused on the duration of IMV, mortality, and complication rates. We used a prospective cohort study design to compare the ‘exposures’ of early (performed at day 7 to 10 after starting IMV) versus late (performed after day 10) tracheotomy and set the treatment decision time on the 7th day after orotracheal intubation.Implications of all the available evidenceThe evidence suggests that tracheotomy within 10 days of starting COVID-19 patients on mechanical ventilation allows these patients to be removed from ventilation and discharged from ICU quicker than later tracheotomy, without added complications or increased mortality. This evidence may help to release ventilators and ICU beds more quickly during the pandemic.