Minimal clinically important difference for chest discomfort in patients undergoing pleural interventions.

Pleural effusion is a common diagnosis, making thoracentesis one of the most commonly performed medical procedures. 1 2 Large-volume thoracentesis has been associated with several complications, including chest discomfort. 3 Improvement in breathlessness is an important clinical goal for this procedure. Randomised controlled trials evaluating pleural fluid aspiration commonly use the visual analogue scale (VAS) to measure patient-centred outcomes such as dyspnoea and discomfort. 4–9 A VAS for chest discomfort consists of a 100 mm long line with ‘no discomfort at all’ printed at one end and ‘worst possible discomfort’ at the other; patients are asked to make a mark through the line indicating where their current symptoms fall along that continuum, with the distance in millimetres along the line to the mark representing their numerical VAS score. Importantly, interpretation of data derived from these scales requires a determination of the minimal clinically important difference (MCID), defined as the smallest change in VAS score perceived as clinically significant. This contrasts with the statistical significance, which may not be relevant to patient care. For example, the smallest possible between-group difference in VAS score of 1 mm could be statistically significant if a trial were very large. Knowledge of how much a score must change before a patient perceives the change as meaningful or worthwhile is necessary to interpret whether a reported difference is clinically significant. As a result, MCIDs are vital to power calculations in trials using patient-centred primary outcomes measured by VAS, as enrollment targets are selected to allow detection of a difference between study groups at the threshold of MCID. Various methods for determining the MCID have been described. 10 The most direct method involves anchoring a change in the tool being investigated to another scale that directly inquires as to whether a change of significance has occurred, such as a Likert scale. A previously determined MCID for pain of 13 mm on 100 mm VAS derived from emergency department patients with trauma has been used for all recent studies using 100 mm VAS to measure chest discomfort related to pleural fluid aspiration. 4–6 8–12 We hypothesised that acute traumatic pain may differ sufficiently from the chest discomfort associated with large-volume thoracentesis to result in a different MCID. The primary aim of this study was to determine the MCID for the VAS for chest discomfort in patients undergoing thoracentesis using a Likert scale anchor.

Data analysed for this study were obtained during two randomised controlled trials of therapeutic thoracentesis performed at 10 US academic medical centres. One randomised to manometry-guided versus symptom-guided discontinuation of pleural fluid aspiration while the other randomised to drainage by gravity versus manual aspiration; both were negative trials. 4 5 Both trials enrolled adult patients with symptomatic pleural effusions of estimated volume 0.5 L or greater referred for therapeutic thoracentesis. Patients without free-flowing effusions, those unable to maintain a seated position for the procedure and those unable to provide informed consent for the procedure were excluded. All participants provided written informed consent. Thoracentesis catheters were inserted in standard fashion following instillation of local anaesthesia with 1% lidocaine without epinephrine. The institutional review boards at each participating medical centre approved these trials. 4 5 In both trials, patients were asked to indicate their degree of chest discomfort on VAS before, during and after the procedure. Five minutes after drainage catheter was removed, patients were asked to indicate their overall degree of chest discomfort on VAS from procedure start to that moment, followed by a 7-point Likert scale to indicate the degree to which their chest discomfort changed from procedure start to that moment, with the following options: large or moderate improvement, small but just worthwhile improvement, slight improvement in chest discomfort but not worthwhile, no change, slight increase in discomfort but not significant, small but significant increase in chest discomfort and large or moderate increase in discomfort. See figure 1 for the Likert scale and example of VAS scale used in these trials. Example of visual analogue scale and Likert scale used in these trials. Mean change in VAS scores between preaspiration and 5 min postprocedure was calculated along with interquartile ranges, stratified by response on the 7-point Likert scale. The threshold for MCID was defined by respondents indicating a ‘small but significant increase in chest discomfort’. Descriptive statistics including means and SDs for continuous variables and percentages and frequencies for categorical variables were also performed. All analyses were done with R V.3.3.1. Patient and public involvement The current study used existing data from two randomised trials. Our research question is highly patient centred, by seeking to better characterise the measurement tools used to document discomfort in clinical trials, which will in turn inform future clinical trials of pleural intervention and thus have lasting impact on patient outcomes.

A total of 263 study thoracenteses were performed in the course of the two trials, the first enrolling March 2016 to September 2017 and the second between October 2018 and April 2019. All had complete outcome data, including preprocedure and postprocedure VAS scores and Likert scale responses. Mean age was 65 years (56–72 years), 52% were women and two-thirds had comorbid malignancy. See table 1 for additional demographic information. The most common final effusion diagnosis was malignancy, accounting for 115 effusions (60%), with additional aetiologies listed in table 2 . Mean volume drained was 1150 mL. Most procedures were stopped due to discomfort (112 patients, 43%) or spontaneous cessation of flow after complete effusion evacuation (116 patients, 44%), with remaining discontinuation reasons listed in table 3 . Subject demographics Effusion etiologies Reasons for discontinuation of pleural fluid drainage *Cases allocated to the manometry arm of manometry trial. 6 Thirty-four of 263 patients experienced a ‘small but significant increase’ or a ‘large or moderate increase’ in discomfort following thoracentesis. The mean increase in VAS score in those reporting a ‘small but significant increase’ in chest discomfort (n=23) was 16 mm (SD 22.4, 95% CI 6.9 to 25.2). Table 4 and figure 2 illustrate the trend in change in VAS score by Likert scale responses. Ninety-eight of 263 patients (37%) were using narcotic analgesics at the time of the procedure. Those taking narcotics did not differ in reported a mean overall chest discomfort score (mean difference 4.3 points higher in those on narcotics, 95% CI −1.4 to 10.0, p=0.14). Preprocedure to postprocedure change in visual analogue scale discomfort score according to Likert scale response. Likert Scale responses VAS, visual analogue scale.

Using data from two recent multicentre randomised trials in which all patients completed VAS for discomfort preprocedure and postprocedure as well as postprocedure Likert scales, we established a VAS MCID of 16 mm for chest discomfort related to pleural fluid aspiration using the anchor method. The VAS is widely used in pleural intervention trials as a standardised tool to translate subjective symptoms such as dyspnoea or discomfort into a numerical construct. It therefore plays a prominent role in power calculations for clinical trials related to pleural interventions. Despite its widespread use, the MCID for discomfort specific to pleural interventions has not been previously defined. The discomfort of excessively negative pleural pressure is unlikely to be experienced in the same manner as other forms of pain or discomfort. A wide range of VAS MCIDs for pain and discomfort of different aetiologies have been described, including 10 mm in acute postoperative pain 13 and pain due to endometriosis, 14 13 mm for patients with acute traumatic pain in emergency department, 10 11 18 mm for total hip arthroplasty 15 and 22 mm for total knee arthroplasty. 15 The pleural discomfort MCID of 16 mm falls within this previously described 10–22 mm range for varying kinds of discomfort. From a trial design perspective, however, the number of trial subjects required to have sufficient power to detect a difference of 10 mm between groups tends to be markedly different than that required to detect a difference of 22 mm between groups. Using the power calculation assumptions used in GRAVITAS, for example, power to detect a 10 mm difference would require 282 total subjects versus just 58 to detect a 22 mm difference. 5 Knowledge of the specific VAS MCID for pleural procedures is therefore crucial for well-designed future studies intended to demonstrate significant improvement in this important patient-centred outcome. There has been a dramatic rise in clinical trials in pleural disease in the last decade, the majority of which use patient-centred primary outcomes, such as pain or dyspnoea. Our study now provides a definitive way to interpret and contextualise these subjective outcomes in future trials. This study has several limitations. Although we had a relatively large total sample size of 263 thoracenteses, only 23 patients reported a ‘’mall but significant increase in discomfort’ on Likert scale from which the MCID can be calculated, leading to a relatively wide 95% CI and therefore somewhat less certainty in the precision of the MCID value we report. However, in a similar study which determined the VAS MCID for improvement in dyspnoea related to pleural intervention, only 20 patients registered the Likert category of interest also resulting in relatively wide confidence intervals 7 but has nonetheless been used widely in the development of subsequent trials. Other prior VAS MCID study populations have demonstrated similar findings. 10 Another potential limitation of this work is that it only included thoracenteses but not other pleural interventions such as tube thoracostomy. However, these procedures tend to be similar in technique and catheter size, suggesting these findings may generalise to all small-bore pleural catheter procedures. While more invasive pleural procedures, including medical thoracoscopy and large-bore chest tube placement, may result in different discomfort patterns than that described by the MCID reported here, our study provides more relevant information than previously published non-thoracic studies. Finally, the clinical trial population used for this study may not be fully generalisable to non-study-based patient populations. In conclusion, this study determined the MCID on VAS specific to chest discomfort during pleural drainage procedures is 16 mm. Chest discomfort is an important patient-centred clinical outcome of pleural procedures; this specific MCID can be used to plan and interpret results of future pleural intervention trials.