Tonsillectomy

Tonsillectomy – Who and
Why?
Janet A Wilson
BAPO
2011
MacKee 1963 n=413 ??entry + outcome
criteria; exclusions from analysis
Mawson 1967 n=404 ?outcomes
Roydhouse 1970 n=426 - allocation, observer
and recall bias
Paradise 1984 n= 91, more severely affected
( 7 in 1 y 5 in 2y) small effect
Paradise less affected (2 - 6 in 1y, 2-4 in 2y)
pub’d eventually in 2002 – even smaller effect
Cochrane review Burton 2007 ; Blakley 2009
OHNS– systematic review; Schilder 2004 BMJ
Conclusion – tonsillectomy works a wee bit.
1998 International Paed T’s rate
per 10,000 population (Akker, 2004)
Canada
Scotland
USA
Australia
N Ireland
0
20
40
60
80
100
120
140
15 year fall in rates of UK ENT
surgery
80000
70000
60000
50000
Adenoids
Grommets
Tonsils
40000
30000
20000
10000
0
Mid 1990s
2008 - 09
UK Tonsil Admissions
40000
1990-00
35000
2008-09
30000
25000
20000
15000
10000
5000
0
Child Op
Child Emergency
NHS Tonsil Ops <15years old per 100,000
population Suleman Clin Otolaryngol 2010
800
700
600
500
400
300
200
100
0
Barrow
Tameside
Calderdale / Luton
Oldham
Walsall
Manchester
Kensington
Harrow
Rushcliffe
729 eligible; 268 randomised
NESSTAC Arch Dis Child 2010
70
Medical
60
RCT
50
40
Surgery
30
20
10
0
n sore
throats
%
Mean days % School
Persistent off school progress
s/t
down
ITT comparison of the NESSTAC
randomised arms.
Tonsillectomy: fewer episodes of sore
throat
Relative Risk Year 1 = 0.74; Year 2 = 0.51
The estimated Intention to Treat effect of
tonsillectomy in randomised patients = 3.5
fewer sore throat episodes in two years
Incremental cost-effectiveness ratio of
£261 per sore throat avoided.
% Switch rates – 36% overall
30
25
20
15
10
5
0
RCT Med
RCT Surg
Cohort Med
Cohort surg
The “average” child
Episodes of sore throat saved
9
8.7
8.4
Year 1
8.1
8
7.1
7
6
n
Year 2
Total
5
4
3
2
1
0
1
2
3
6
Months delay
As treated analysis
As-treated analysis = > 8 sore throats
saved in 2 years
if operation is performed <10 weeks of
consultation
= £154 per sore throat avoided.
Quality of Life
T14 –Hopkins et al JLO 2010 – effect size =
1.53
PedsQL – NESSTAC – similar physical health at
24 months –surgical preference group worse at
baseline (HTA Journal 2010; 14, No. 13 )
OHNS Supplement 2008.
– Goldstein - improvements general and disease
specific tools
Lock J of Child Health Care 2010 – emphasises
the family QoL impact
NESSTAC Conclusions
Tonsillectomy significant clinical effect in reducing
sore throats, at a reasonable cost.
The effect depends on timing of surgery, as the health
of all children with recurrent sore throat improves over
time.
Qualifying children should be treated promptly, as
early intervention = maximum benefit.
This is important if there is a high social or
educational impact.
Early intervention – cost £154 per sore throat saved
More baseline data would have informed the analysis
Core problem of surgical trials
“Intention To Treat” RCT analysis
But worst cases tend to switch from the nontreatment arm into treatment
Surgery is irreversible: issues of acceptability;
design ( no cross over)
Net result = modest effect = easily dismissed
as “Expectancy bias”. i.e. surgical placebo
component (essentially uncontrolled for)
No shift in surgical criteria or perception of
effectiveness
Willingness to pay
Burton – “how much uncertainly are HC
providers willing to fund?”
How much were NESSTAC parents willing
to pay for tonsillectomy?
– Mean = £8000
– Median = £5000
http://www.hta.ac.uk/pdfexecs/summ1413.pdf
Clinicians who perform
tonsillectomy should
determine their rate of
primary and secondary
post tonsillectomy
haemorrhage at least
annually
Watson Newcastle upon Tyne
1993
N=1036
Monopolar diathermy faster than ties
No difference 2° bleeds
Study powered e.g. to detect difference
3% versus 7% secondary haemorrhage
Bipolar Diathermy dissection
1994: Descriptive; 4 ml blood loss.
2° haemorrhage = 3%
1995: randomised n=120
BP faster (11min) and less blood loss (5
ml)
2° Haemorrhage –
3.4% cold (2 pts) vs. 1.7 % BP diathermy
Needed ~ 3000 pts to achieve significance
BP versus Cold in Portugal Silveira
2003
N=60; alternating allocation
Needed 34% bleed in one group for
significance
BP vs Cold Turkey Kazali 2005
N=40, randomised
BP faster (14min) and less blood loss (14
ml)
2° Haemorrhage – Nil…
Powered to detect 0% versus 38%
UK Audit (2005) n=33,900
www.rcseng.ac.uk
90% power to detect significant
differences with p<0.01……
3.9% readmission
OR haemorrhage BP diathermy = 2.47
BP haemostasis not dissection = 1.57
Laryngoscope 2010 Tomkinson et al
Welsh Data 2003-08
N=17480
All techniques using heat had a sig.
increase in secondary haemorrhage
Odds of Secondary Haemorrhage
14
12
10
8
6
4
2
0
Cold D
Diss
+BP+tie
Diss+BP
Coblation
Cobl+BP
or ties
What IS the bleeding rate?!
Secondary bleeds….
Takwoingi 2007 JLO UK 8-9%
Attner 2009 Sweden - Karolinska - 5.5%
Gi 2010 Netherlands – Guillotine 0.28%
Kim 2010 Korea 3% children; 10% adults
Mortality – recent data
German Surveillance unit for rare paed
disorders – 2 deaths in 2 years –
Bidlingmaier 2010
32 deaths identified in literature Windfuhr
2009
Suggest about one per annum in UK?
Most deaths due to airway obstruction
Clot is hard to get out
MUST have capacity for rigid
bronchoscopy – STAT - plus or minus
trache
Surgical Standards (Narula et al Clin
Otol 2011)
“% surgeons operating on children who
are trained in Paediatric Basic Life
Support”
46%
3 hospitals only met one standard
Sore throat Take Home messages
We need more baseline data about ALL sore throats,
and their response to surgery. ? BAPO can set up a
national primary / secondary care interface sore
throat pathway (better than waiting at the ‘end of the
line’ to offer an operation)
We have reached the limits of RCTs
Most bleed rate studies = massively underpowered
Secondary haemorrhage is sig > with ‘hot’
techniques and proportional to the amount of ‘heat’
used, peaking with Coblation plus other methods.
What will it take to convince ENT surgeons if audits
of tens of thousands of patients fail?
UK has ~ one tonsillectomy death per year
Resuscitation skills seem woefully maintained – at
least in London – time for urgent remedial action!
Considerations in T+A for OSA
Selection
OSA cure rate
Other outcomes –
– Behaviour
– Neuro-cognition
– Quality of life
Risk
Sources
AAO HNS Guidelines Roland OHNS 2011
BMJ Clinical Review Powell 2010
UK working party Robb 2009
Meta-analysis in obese children Costa
OHNS 2009
Beyond tonsillectomy Praud 2008
Review – Behavior, neurocognition and
quality of life – Mitchell IJPO 2006
? WHO ?
25% fail – at least
Sleep study –
– diagnosis
– Severity
– Central apnoea
– Risk of complication (high AHI –Ye, J of
OHNS 2009)
? No brainer? – get a sleep study
….or – perhaps not bother
AAOHNS – sleep study if co morbidity or if
need for surgery is ‘uncertain’
– Admit for monitoring if severe i.e. AHI> 10 /hour
or SaO2 <80%
UK Position – “the normal child with severe
OSA is also at risk of perioperative
problems but is … difficult to identify”
– “children with severe OSA should be referred
for further investigation”
– “it is well recognised that a proportion of
children with severe OSA are acutely sensitive
to ..anaesthetic agents”
BMJ 2010
– “Polysomnography is the optimal
investigation”
– “It is not logistically possible for all snoring
children to have polysomnography”
RCPCH management guideline – “reliable
discrimination of snoring and OSA
requires polysomnography”
Consider
Beyond tonsillectomy Praud 2008
Orthodontic features of OSA – LofstrandTidstrom; Souki both 2010 IJPO
What about BMI? How do YOU assess
child’s weight / BMI in OPD?
Cochrane
Natural Hx not fully delineated
“The quality of research in this area could
be improved by the use of sleep studies at
baseline”
Review – Behavior, neurocognition and
quality of life – Mitchell IJPO 2006 – same
conclusion. More investigation
BASELINE evidence is missing – as it was
in NESSTAC
Challenges for BAPO?
Mandate for up to date resuscitation skills
across the board
Better understanding of the natural history
– Primary and secondary care work– capture the
sore throat ‘prequel’ in real time.
– T14 review of ALL subjects not just the
operative cases - More willingness to pay work
Better handle on emergencies
For OSA: work towards use of pre and
post operative ; better risk assessment
OSA-18 and sleep testing