Background:The psychometric properties of the Headache Impact Test ® (HIT) were compared with the 17-item Short Pain Inventory © (SPI) in 75 primary-care patients presenting with headache. A specialist neurologist made formal diagnoses consistent with the International Headache Society classification. We compared the severity of the headaches and the diagnostic labels with the ability of the SPI and the HIT to discriminate severity and diagnosis.
Aim:To compare and contrast the psychometrics of the HIT and the SPI in headache patients in relation to pain severity and diagnosis.
Methods:The dynamic version of the computerized HIT (HIT-DYNHA ® ) was used via the Internet. Patients were given a total of 7 days' supply of the SPI to use from the start of their next headache and to continue over the following week. Neither the HIT nor the SPI claim to be diagnostic tests, but both claim to be reliable and valid outcome measures of headache. Both tests were compared by screening samples identified by: no pain, mild, moderate, severe or extreme pain. They were further stratified into: chronic daily headache, migraine and acute tension-type headache. This tests the power to resolve small differences in severity for each type of headache.
Results:The HIT correlated 0.283 with the severity of the headaches whereas the SPI correlated 0.768 on Total Pain Disturbance. The correlations of the HIT and the SPI mood scales with pain severity never rose above 0.26 whereas the correlations of the SPI were typically 0.7. The SPI produced highly significant discriminability of the severity intervals (no pain, mild, moderate, severe, extreme) with Student t-test values of 2-20. The HIT scores were far less powerful at discriminating severity with t-test values from 2-5. The SPI has far better discrimination than the HIT at resolving the fine detail of severity. HIT scores at screening produced significant discrimination of the diagnoses with the resolution between tension-type headache and chronic daily headache reaching 1/100 million. The SPI summary scores and subscales did not discriminate between diagnoses. Confirmatory factor analyses were carried out including headache diagnosis, SPI and HIT variables. Various Quartimax and Varimax analyses strikingly gave the same two factors of severity and diagnosis. The first factor's major correlation loading was the SPI 'total pain disturbance' (0.97) and loading highly on all SPI subscales (0.85-0.92) and severity (0.77). The HIT loading was -0.17 and diagnosis loaded 0.16. The second factor was identified and loaded as 'diagnosis' (0.94) with HIT loading -0.80. The SPI and severity subscales were correlation loaded 0-0.12 and severity 0.2. The HIT and the SPI clearly measure different things.
Conclusion:The HIT was poorly related to the severity of the pain but very closely related to the diagnostic label. In contrast, the SPI was very closely related to pain severity but not related to diagnoses.