Abstract
In this issue of Cephalalgia, Lucas et al. report on the natural history of headache following mild traumatic brain injury (TBI) in a civilian patient population. This interesting study adds to the growing body of knowledge related to the prognosis and characteristics of post-traumatic headache (PTH). It is also one of only a few studies on head-injured patients that classified headaches using standardized diagnostic criteria over multiple time points, so it is of particular interest.
Study subjects consisted of 212 patients (76% male, mean age 44) at one level 1 trauma center. Subjects were enrolled within one week of sustaining a mild TBI (concussion), most often due to a motor vehicle accident (58%) or fall (24%). Interviews were initially conducted face to face and then by telephone up to one year after the injury.
Subjects with “new or worse” headaches compared to before the injury were considered to have PTH. Headaches were classified as migraine, probable migraine, tension-type, and cervicogenic. Previous studies by this team (1,2) were based on a cohort of Veterans Administration (VA) rehabilitation patients with a wider range of head injury severity although used similar headache classification methodology. The primary findings of the present study were that a) between 54% and 69% of subjects reported new/worse headaches at each interview although a smaller number (41%) reported this consistently at all interviews; b) of those reporting headache, the predominant subtype at three months was tension-type (37%) followed by probable migraine (25%) and migraine (24%); and c) peak incidence of new/worse headache was at the baseline (< 1 week post-injury) and first follow-up (three months post-injury) interviews. Several questions naturally flow from this and other recent studies.
1) What is the expected incidence of PTH and to what extent can we attribute this to the injury per se? In the present study, new-onset headache in general and migraine-like headache in particular were clearly more likely in these patients with head injuries than would be seen in an age- and gender-equivalent nonclinical population. For example, the one-year incidence of International Classification of Headache Disorders (ICHD) migraine in men in the general population has been reported to be less than 1 per 100 person years (3) although the incidence of subjectively worsened headache is unknown. What would be the expected incidence of ICHD migraine in a comparable “at-risk” population such as (for the present study) a predominantly male inpatient population with recent injuries of comparable severity not involving the head? It is possible, even likely, that an injured control group without head injuries would have a higher than expected incidence of headache for a variety of plausible reasons. The reality is that it is difficult to find truly “comparably injured” control groups without head injuries and then there is the added issue of generalizability and overmatching even when available. The few (to our knowledge) prospective studies of TBI patients and injured controls (4–7) were not consistent regarding whether headache was more common in TBI patients vs injured controls.
What is the PTH phenotype? The PTH phenotype, if there is one, is still uncertain and may be different for civilian vs combat-related (predominantly blast) injuries (8). Migraine/probable migraine was the most common headache type in the present study (49% of headaches). Migraine-like headache was also the predominant headache type in other recent studies of civilian and military/VA headache populations that used ICHD criteria (1,9–12) although many earlier studies favored tension-type headache as the predominant PTH phenotype (13,14). Clinically, it is not uncommon for patients to discount nondisabling, remote, or infrequent headaches. This type of reporting bias might be expected to be an ever greater challenge when conducting phenotypic studies in nonclinical populations.
We note that even if PTHs are not obviously different from primary headaches using the relatively crude ICHD classification, more refined analysis of specific ICHD headache features (photophobia and aura being obvious examples) and novel features may reveal potential diagnostic markers in future epidemiological studies. Perhaps further examination of the unclassified group may yield distinct descriptions of PTH worthy of classification.
2) What is the prognosis of PTH? Some interesting findings in this study relate to PTH prognosis. While almost all (92%) patients reported new/worse headache at one or more timepoints, the prevalence of persistent headache (headache at all three interviews) was 41%. Without more details it is difficult to determine whether this is reflective of delayed onset of headache in some patients, high rates of remission, natural variability, or some or all of the above. The Walker longitudinal study (11) suggested high rates of both headache incidence and remission in the year following head injury, which resulted in a somewhat constant prevalence over this one-year period. As the present study and other ongoing longitudinal studies continue to collect follow-up data, it is hoped that we will be able to identify, empirically, the subgroup of head-injured patients who are most likely to develop persistent problematic headaches and ultimately provide more useful prognostic data for these patients. The contributory role of comorbidities (e.g. psychosocial factors, nonheadache pain, or other injuries) might be important and should be taken into account in estimates of PTH prognosis.
3) Is it time to ditch the seven-day rule? The authors and others (2,9,15) argue that requiring headaches to manifest within one week of head injury will miss perhaps 20% to 30% of PTH patients. In our opinion there is still insufficient evidence to recommend a specific alternate cut-off point (one month? three months?) but there is enough uncertainty that it makes sense to capture the date of onset in ongoing observational and interventional studies so that results will be “upward compatible” if this criterion is relaxed in the future. Further studies are necessary to ascertain whether immediate vs delayed onset PTH differ.
Limitations of this study are of course the lack of a control group and whether findings generalize to less severely injured subjects or patients with different types of injuries. It is noteworthy that while the head injuries in these patients were mild, other injuries were of sufficient severity to result in hospitalization in most cases. The prevalence of pre-injury headache was only 17%, which seems low although appears to include only “problematic” headaches.
According to this and other recent studies, PTH may be more disabling than previously recognized. Well-controlled epidemiologic studies with determinants of headache persistence are necessary to adequately assess headache prognosis. A classification system that subtypes PTH not only into primary headache subtypes but also by specific features may be desirable in future observational and interventional research even if not of immediate clinical relevance.
Footnotes
Conflict of interest
None declared.