Professional Documents
Culture Documents
This appendix has been provided by the authors to give readers additional information about their work.
Supplement to: Brown SGA, Ball EL, Perrin K, et al. Conservative versus interventional treatment for spontaneous
pneumothorax. N Engl J Med 2020;382:405-15. DOI: 10.1056/NEJMoa1910775
SUPPLEMENTARY APPENDIX
1
Centre for Clinical Research in Emergency Medicine, Harry Perkins Institute of Medical Research,
Royal Perth Hospital and the University of Western Australia, Perth, WA, Australia
2
Aeromedical and Retrieval Services, Ambulance Tasmania, Hobart, Tasmania, Australia
3
Department of Respiratory Medicine, Royal Hobart Hospital, Hobart, Tasmania, Australia
4
Medical Research Institute of New Zealand, Wellington, New Zealand
5
Capital and Coast District Health Board, Wellington, New Zealand
6
Emergency Department, St George Hospital, Kogarah, NSW, Australia
7
St George Clinical School, Faculty of Medicine, University of New South Wales, Kensington, NSW,
Australia
8
Emergency Department, Monash Medical Centre, Clayton, VIC, Australia
9
Department of Medicine, School of Clinical Sciences at Monash Health, Clayton, VIC, Australia
10
Adult Emergency Department, Park Road, Grafton, Auckland 1142, New Zealand
11
Emergency Department, Gold Coast Health Service District, QLD, Australia
12
School of Medicine, Bond University, Gold Coast, QLD, Australia
13
School of Medicine, Griffith University, Gold Coast, QLD, Australia
14
Emergency Medical and Children's Services, The Prince Charles Hospital, QLD, Australia
15
University of Queensland, QLD, Australia
16
Department of Respiratory and Sleep Medicine, The Sutherland Hospital, Sydney, NSW, Australia
17
Royal Perth Hospital Imaging, Royal Perth Hospital, WA, Australia
18
Respiratory Medicine, Sir Charles Gairdner Hospital, Perth, WA, Australia
19
Centre for Respiratory Health, School of Medicine & Pharmacology, University of Western Australia,
Perth, WA, Australia
20
Pacific Radiology, Wellington, New Zealand
21
Department of Respiratory Medicine, The Cairns Hospital, Cairns, QLD, Australia
22
Department of Cardiothoracic Surgery, Monash Health, Clayton, VIC, Australia
23
Department of Surgery, School of Clinical Sciences at Monash Health, Monash University, Clayton,
VIC, Australia
24
Respiratory Department, Royal Perth Hospital, Perth, WA, Australia
Figure S4: Intention To Treat (ITT) Kaplan-Meier survival plot for time to radiographic 11
pneumothorax resolution.
Figure S5: Intention To Treat (ITT) Kaplan-Meier survival plot for time to symptomatic recovery, i.e. 12
last symptoms or use of medication to treat symptoms related to pneumothorax.
Figure S6: Intention To Treat (ITT) Kaplan-Meier survival plot for pneumothorax recurrence. 13
Table S1. Participant characteristics 14
Table S2. Reasons for conversion to invasive management in the conservative management group 15
Table S3. Agreement between treating clinician and masked radiologists for CXR 16
resolution at 56 days (for CXRs available for masked assessment by radiologists)
Table S4. Study-related procedures by treatment group 17
Table S5. Length of stay in the first eight weeks, by treatment group 18
Table S6. Radiological procedures by treatment group 19
Table S7. Patient-related outcomes and satisfaction at final assessment by treatment group 20
Table S8. Interaction analyses with pre-specified variables 21
Victoria, Australia:
Box Hill Hospital, Box Hill, Vic, Australia (Paul Buntine, Andrew Maclean, Julia Ng); Casey Hospital,
Berwick, Vic, Australia; Dandenong Hospital, Dandenong; Vic, Australia; Monash Medical Centre,
Clayton, Vic, Australia (Ali Asadpour, Gaby Blecher, Simon Craig, Diana Egerton-Warburton, Andis
Graudins, Barton Jennings, Robert Meek, Alastair Meyer, Kirsty Povey, Rachel Rosler, Julian Smith,
Kathryn Wilson)
Tasmania, Australia:
Royal Hobart Hospital, Hobart, Tasmania, Australia (Emma Ball, Simon Brown, Geoffrey Couser, John
Dewing)
Queensland, Australia:
Bundaberg Base Hospital, Bundaberg, Qld, Australia (Pradeep Bambery, Michael Chang, Greg
Treston); Gold Coast Health Service District, Southport, Qld, Australia (Gerben Keijzers, Toby Tang);
Ipswich Hospital, Ipswich, Qld, Australia (Kylie Baker, Adel Braasch); Logan Hospital, Meadowbrook,
Qld, Australia (Deepak Doshi); Mater Hospital, South Brisbane, Qld, Australia (Simon Bowler, David
Serisier, Joseph Ting); Nambour General Hospital, Nambour, Qld, Australia (Michael Bint, John Fuller,
Ogilvie Thom, Yusuke Ueno-Dewhirst); Royal Brisbane and Women's Hospital, Herston, Qld, Australia
(Kevin Chu, Duncan McAuley, Christopher Zappala); The Cairns Hospital, Cairns, Qld, Australia (Mark
Little, Graham Simpson, Stephen Vincent); The Prince Charles Hospital, Chermside, Qld, Australia
(Frances Kinnear, Philip Masel); The Townsville Hospital, Douglas, Qld, Australia (Jeremy Furyk,
Huang-Liang Lee, Anthony Matthieson); Toowoomba Hospital, South Toowoomba, Qld, Australia
(Simon Tebbutt, Kathleen Hyland, Ross Sellars)
New Zealand:
Auckland City Hospital, Auckland, NZ (Peter Jones, Margaret Wilsher); Christchurch Hospital,
Christchurch, NZ (Lutz Beckert); Middlemore Hospital, Auckland, NZ (Jeff Garrett, Hamish Read)
Waikato Hospital, Hamilton, NZ (Catherina L Chang, Hollie Ellis, Robert J Hancox); Wellington
Hospital, Wellington, NZ (George Bardsley, Richard Beasley, Kyle Perrin, Sharon Power)
The primary outcome was assessed by an absolute risk difference and confidence interval with a non-
inferiority bound of -9%. A non-significant P-value implies that the null hypothesis of inferiority is
accepted for conservative compared to interventional therapy. A significant P-value means the null
hypothesis of inferiority is rejected in favor of the alternative hypothesis of non-inferiority.
Interactions between the primary outcome variable by intention to treat were assessed for age,
clinician estimated size, and symptom duration (illustrated by comparison of odds ratio for association
comparing the 25th and 75th percentiles of these variables from the control group data summaries);
and by smoking status. A logistic regression model was used. An interaction P-value that was not
statistically significant means there was no difference in the effect of randomized treatment in relation
to the sub-group.
McNemar’s test for paired proportions and estimation of the paired differences for clinician compared
to radiologist for ‘not resolved’ are used for assessment of agreement
Relative risk for an event and absolute risk difference expressed as a percentage were estimated for:
Adverse events, pneumothorax recurrence by 12 months, pneumothorax recurrence overall, use of a
drain for 72 hours or more, use of at least one CT scan, use of at least one invasive procedure, and
complete symptomatic recovery.
The Mann-Whitney test and the Hodges-Lehmann estimator of location shift with 95% CI were used
for Hospital bed-days, count of CT scans, and days off work due to pneumothorax.
A t-test was used to estimate the difference in the mean number of CXRs. Poisson regression was
also used for the rate of use of CXR, CT scans, number of adverse events, and invasive procedures.
Ordinal regression was used to estimate the odds of a higher (better satisfaction) versus lower score
(worse satisfaction) on the satisfaction index. An odds ratio of greater than one favors the second-
named therapy.
SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) was used for all analyses.
The Statistical Analysis Plan did not specify the window for an eight week visit, nor how missing data
by Day 56 would be handled. Analysis was undertaken on those participants who had data available
by this designated time point. As a result, we undertook a complete-case-analysis in which data from
participants in whom the eight week clinic visit occurred after eight weeks were treated as ‘missing’,
unless a CXR during the eight week window or later demonstrates a persisting pneumothorax thereby
confirming treatment failure. The ‘window’ for an eight week visit outcome was the week up until 56
days; thus any demonstrated resolution prior to 56 days was taken as a primary outcome of resolution,
any persistent pneumothorax after 49 days (including after 56 days) was taken as a primary outcome
of failure, and remaining cases were treated as missing data.
In the primary sensitivity analysis the time point was extended to Day 63 (ie eight weeks plus 7 days).
The rationale for this analysis was that the strict 8 week time point for the primary outcome variable
set at day 56 did not allow inclusion of data from participants in whom their eight week clinic was
scheduled for, and/or took place after 56 days. The intent of the study was to establish whether there
was resolution at the time of the week eight visit. The protocol did not specify that the eight week
follow-up must be undertaken before 56 days, and investigators were not advised there was a strict
cut-off for interpretation of the eight week follow-up CXR. Nor did the protocol state a ‘window’ in
which the eight week visit should take place. The frequency plot (Figure S1) provides information of
the distribution of week eight visits around the Day 56 time point.
In the secondary sensitivity analysis, missing data at Day 56 was imputed as ‘failure’ to provide an
estimate of the difference between groups, if all missing data represented actual treatment failures
using a strict Day 56 cut-off.
Figure modified from Collins et al. Quantification of Pneumothorax Size on Chest Radiographs Using
lnterpleural Distances: Regression Analysis Based on Volume Measurements from Helical CT.
American Journal of Roentgenology. 1995;165:1127-130.
Participants with CXRs showing non-resolution of pneumothorax in this time period are considered as
treatment failures in the primary analysis, and not included in this figure.
management group
(Participants may have had interventional management for more than one reason)
Intolerable symptoms 8
Difficulty mobilizing 3
Hemothorax 2
Patient anxiety 2
Slow resolution 1
Radiologist resolved
No 2 0 2 (0.8%)
Yes 14 223
Chi-square DF P
Suction application only in those who 51* 6.5 5 0 to 18.6 11* 4.7 3.9 0.2 to 10.7
received suction (Days) (5) (2.1 to 10.6) (3.3) (1.8 to 7.7)
* Drain end time not available for one participant in each group, suction end time not available for one participant in each group
VATS: Video-assisted thoracic surgery
Total Length of stay (Days) 154 6.1 3.8 0 to 47.7 162 1.6 0.2 0 to 28.2
(7.6) (0.8 to 9.3) (3.5) (0.2 to 0.8)
- Initial Length of stay 149 4.6 3 0 to 39 161 1 0.2 0 to 28.2
(Days) (5.7) (0.7 to 6.1) (2.8) (0.2 to 0.6)