### Prognosis Scenario

You are a health visitor doing a Child Health Clinic and are approached by the mother of a 2 year old boy who has glue ear, manifest by intermittent ear discharge and hearing difficulty. The mother has seen the ENT surgeon who decided against surgery. She wants to know whether glue ear might cause learning and behavioural problems later in childhood.

What do you tell her?

She asks whether you think she should pay for private grommet surgery. What is your advice? You pose the question, are children with middle ear disease at increased risk of behavioural problems? You recall seeing an article about this in a recent issue of Archives of Disease in Childhood. You review recent issues and find it.

Read the article and decide:

• Is this evidence about prognosis valid?
• Is this valid evidence about prognosis important?
• Can you apply this valid and important evidence about prognosis in caring for your patient?

### Completed Prognosis Worksheet for Child Health

#### Clinical Question

Would grommet surgery reduce the likelihood of behavioural problems later in childhood?

#### Are the results of this prognosis study valid?

Was a defined, representative sample of patients assembled at a common (usually early) point in the course of their disease?
Yes. Large birth cohort of more than 12,000.
Was patient follow-up sufficiently long and complete?
Maybe. Follow up was for 10 years. Original number stated as more than 12,000, at 5 years about 12,000 and at 10 years about 9,000. Authors comment that the prevalance of middle ear disease was similar in children followed as in those lost to follow up but no figures given.
Were objective outcome criteria applied in a “blind” fashion?
Yes. Scales for behaviour and cognitive ability based on parent and teacher responses to questions. Respondents were not necessarily blind to middle ear disease but many factors were studied, and therefore, unlikely to be affected.
If subgroups with different prognoses are identified, was there adjustment for important prognostic factors?
Yes. Adjustment for social class and maternal malaise, but may be incomplete. No other social or cultural factors, such as maternal age or education, taken into account.
Was there validation in an independent group (“test-set”) of patients?
No, but previous cohort studies have reported a possible association.

#### Are the valid results of this prognosis study important?

How likely are the outcomes over time?
Look at risk of neurotic behaviour score greater than 90th percentile at 10 years for children with hearing difficulty based on parent reports (table 9). OR= 1.40 and for teacher reported behaviour (table 10) OR= 1.25.
However, using continuous variables, the largest difference was for antisocial behaviour, with a 0.13 SD difference at 10 years. This is small, and unlikely to be clinically important.
How precise are the prognostic estimates?
Can look at 95% confidence intervals around the odds ratio. CI around OR does not include 1 so result is statistically significant

begin{align}
qquadtext{$OR$ for Hearing Difficulty} &=1.38 (1.1 text{ to } 1.73)\\
qquad text{$OR$ for Ear discharge}&=1.42 (1.17text{ to }1.73)
end{align}

#### Can you apply this valid, important evidence about prognosis in caring for your patient?

Were the study patients similar to your own?
Yes.
Will this evidence make a clinically important impact on your conclusions about what to offer or tell your patient?
Yes. Magnitude of effect is small but there is an association. Not enough information to discuss surgery as no evidence that intervention will alter long term risks.

#### Additional Notes

• Health visitor would probably advise mother to inform school when her son has ear discharge or hearing difficulty, so teachers make an extra effort.
• Surgery has not proven to reduce the behavioural or cognitive problems associated with glue ear, so cost of private surgery probably not warranted without more evidence.
• The small difference in behaviour may nevertheless be important at a population level.
• There is some suggestion of an interaction between social class and the effects of glue ear on behavioural and cognitive problems.

### Middle ear disease – Uncertain whether it increases the risk of behaviour problems at 10 years

#### Clinical Bottom Line

Middle ear disease is associated with a small increase in behaviour problems at 10 years but the difference may not be clinically significant and may be partly explained by socio-economic factors.

#### Citation

Bennett, K.E. Haggard M.P.Behaviour and cognitive outcomes from middle ear disease.
Arch Dis Child 1999;80:28-35 .

#### Clinical Question

Are children with middle ear disease at increased risk of behaviour problems?

#### The Study

• Large, national birth cohort of 12,000 children.
• Middle ear disease (discharge or hearing difficulty) and behaviour measured at 5 and 10 years by parental responses to questions and child assessment. Behaviour at 10 years also measured by teacher responses.
• Follow up of about 12, 000 at 5 years and about 9,000 at 10 years. Authors state that loss to follow up did not significantly affect proportions that had middle ear disease.

#### The Evidence

Difference in behaviour score (measured in SD units) for children with reported hearing difficultly (adjusted for socio-economic status, 95% confidence interval)
Antisocial Neurotic Hyperactive Poor Conduct
Difference in Behaviour Score @ 5 years 0.12
(0.06 to 0.18)
0.20
(0.14 to 0.25)
0.19
(0.12 to 0.25)
0.07
(0.01 to 0.13)
Difference in Behaviour Score @ 10 years -0.01
(-0.07 to 0.06)
0.15
(0.08 to 0.22)
0.25
(0.18 to 0.32)
0.07
(0.00 to 0.14)

#### Comments

• The differences in behaviour score are small. However, middle ear disease is common (10% in this cohort) and at a population level, these small differences may be clinically important.
• The slightly higher adverse behaviour scores in children with hearing difficulty could be partly explained by residual confounding due to socio-economic factors.
• Occurence of middle ear disease was based on parental reports and misclassification of exposure was likely. This may reduce the strength of true associations.