Development and validation of a chronic headache classification interview - CHRONIC HEADACHE EDUCATION AND SELF-MANAGEMENT STUDY (CHESS).

Problem

Many patients with chronic headaches do not have an accurate diagnosis and can receive inappropriate drug treatment and management of their headaches. There are deceptively simple diagnostic criteria for different headache types; for example, NICE headache guidance; in reality, it can be challenging for a non-expert clinician to make an accurate diagnosis. As part of the CHESS study we wanted to develop and validate a telephone classification interview that can be used by a non-headache expert to classify common chronic headache types in primary care and that can specifically:

• Exclude serious pathology (secondary headaches other than medication overuse)

• Exclude primary headache disorders other than migraine and tension type headache

• Distinguish between chronic migraine and chronic tension type headache

• Identify medication overuse headache

Approach

The development of the classification interview was informed by a systematic review of evidence on diagnostic tools for headache disorders; and a chronic headache classification consensus conference held at University of Warwick in October 2015. Conference delegates were randomly allocated to multidisciplinary groups and a nominal group technique used to reach consensus on the key questions to be included in the classification interview.

We trained research nurses to conduct telephone classification interviews with participants recruited from general practice. Doctors specialising in headache conducted a second ‘gold standard’ validation interview with each participant. We measured level of agreement using proportion of concordance, simple kappa and prevalence-adjusted bias-adjusted kappa (PABAK).

Findings

In total 26 delegates attended the headache classification conference: five headache specialist nurses, 13 neurologists (10 with a specialist interest in headache), seven lay representatives and one GP with a specialist interest in headache. We used the results of the meeting to develop a logic model which underpins the telephone classification interview.

We trained six research nurses to use the logic model and complete the telephone classification interviews. Four doctors specialising in headache conducted a second validation telephone interview.

We completed 100 paired interviews, median days between interviews was 32 days (range, 7 to 94). Proportion of concordance was 0.91, the simple kappa coefficient was 0.56 (95% CI, 0.30 to 0.82), the maximum attainable kappa was 0.85 and the PABAK was 0.82 (95% CI, 0.71 to 0.93), a very good agreement.

Consequences

We have developed and validated a telephone classification interview that can be used by a non-expert clinician to classify common chronic headache types, and support GPs and other non-headache specialists to diagnose headaches more accurately and provide more targeted treatment and advice.