Chronic Headache

To validate the classification interview we did paired telephone interviews with participants recruited from primary care as part of a feasibility study for CHESS. The feasibility study received Ethics approval from West Midlands – Black Country Research Ethics Committee (15/WM/0165).
We identified participants from searches of general practice records from 14 practices in the West Midlands region of the UK that cover urban, small town and semi-rural areas with varying levels of deprivation and ethnic diversity. We searched for people who had consulted within the previous year with headache, or had been prescribed migraine specific medication. Due to the typically imprecise coding of chronic headache in primary care and the fluctuating nature of headache frequency the search included people with both episodic and chronic headaches. GPs screened lists of participants identified from the searches and excluded those with known serious underlying pathology or secondary causes of headache (other than medication overuse headache) or terminal illness because we did not want to cause unnecessary upset or distress inviting them to take part in a study of a self-management intervention for chronic headache. People interested in the study were contacted by a member of the study team to confirm that they had experienced headaches for at least half the days of the month and for at least three months. People who met these criteria and provided written consent were invited to take part in two telephone interviews, the first a telephone classification interview conducted by specially trained nurses and the second a validation interview conducted by doctors working for the National Migraine Centre. The nurse interviews were audio recorded for quality assurance purposes.

As part of the ‘Behavioural and Environmental Assessment’ phase, surveys were conducted in 60 villages between February 2011 and January 2012. Since the main study is a community-based randomized control trial, the goal of the sampling was to compare the prevalence of active cysticercosis and the knowledge, attitudes and practices (KAP) relating to cysticercosis and taeniasis among participants through time. Therefore, the responses from the 80 participants per village are meant to represent KAP among those participants and are not meant to be generalized. However, we believe that the data represent the KAP pertaining to cysticercosis among people living in the 60 sampled villages and can be used to develop educational material for these communities. Such educational materials can be adapted to other endemic communities in the country.
Briefly, in each study province, all departments where pigs were raised (30 of 31 departments) were selected, and two villages meeting the eligibility criteria were randomly selected from each department for a total of 60 participating villages. Eligible villages had a population of at least 1 000 people as according to the 2006 census, were on the map provided by the Burkina Faso Geographic Institute (2000), were at least 5 km away from another participating village, were not located on a national or provincial road, were not the capital of the region or of the province, and were not located within 20 km of Koudougou or Ouagadougou. In each of the participating villages, 80 concessions were selected (a grouping of several households, usually members of the same family), including at least ten concessions with sows and at least 30 with pigs aged less than 12 months. One household was selected at random from each concession. In each sampled household, one eligible member was randomly selected to participate in a 3-year follow-up study on epilepsy and chronic headaches. As part of the follow-up study, participants had to answer a baseline questionnaire that was included in the present study (see Additional file 2). The inclusion criteria for participants in the baseline study and subsequent follow-up study were: aged at least 5 years, and having lived in the village for at least 12 months and not planning to move in the next 3 years. Figure 1 shows the location of villages used in the pilot study [4 (link)–6 (link)], the GDs, and the surveys.Fig. 1

Villages used in the pilot study, GDs, and questionnaire survey during the planning of the health education intervention strategy to control taeniasis and cysticercosis in Burkina Faso, 2007–2012. One village (green star) was involved in both the pilot and the main study

The surveys were largely based on a cysticercosis questionnaire developed by the Cysticercosis Working Group in Eastern and Southern Africa (CWGESA). They measured sociodemographic factors, behaviors related to pork consumption, and knowledge on porcine cysticercosis, taeniasis, and of the life cycle of T. solium. In the first 40 selected concessions with sows or piglets, a blood sample was taken from one pig to measure the presence of active infection with cysticercosis. The owner of the sampled pig was asked to answer a questionnaire regarding pig management and knowledge on porcine cysticercosis. All data were recorded on personal digital assistants (HP iPAQ Rx5910 Travel Companion, USA). These electronic devices allowed better quality control during data collection and prevented potential errors during data entry.

Participants were dichotomized into those with chronic headaches (CH) and those with no such history (NH). Patients with physician-diagnosed and managed post-traumatic headaches and migraines were all included in the “chronic headaches” cohort. Though the mechanisms underlying the development of these two diseases are different, definitive clinical diagnosis one way or another can be difficult [26 (link), 27 (link)], especially among pediatric athletes involved in high contact sports. This difficulty can be attributed to the significant phenotypic overlap between the two conditions, including throbbing headache, pain on the side of the head, nausea, vomiting, and headache exacerbated by activity [28 (link)]. Additionally, pediatric patients report their symptoms and history of present illness less reliably, further increasing the difficulty of definitive diagnosis. Patients with chronic headaches of unspecified classification were also included in the analysis. In the clinical setting these cases should be followed longitudinally to determine and confirm an exact diagnosis. Given our study design, we lacked this privilege. However, excluding these cases would provide only a partial account of the chronic headache burden among the pediatric athlete sub-population [4 (link)]. Appropriate sample sizes for each cohort were derived using the equation below. $$n=\frac{{[z_{\alpha }\sqrt{2PQ}+z_{\beta }\sqrt{p_1{q}_1+p_2{q}_2}]}^2}{{(p_1-p_2)}^2},whereP=\frac{x_1+x_2}{n_1+n_2}andQ=1-P$$
According to the power analysis, we needed at least 524 patients in each group to detect a mean concussion incidence difference of 10% between the CH and NH cohorts, at 80% power and with a Type I error set at 0.05.
Chi-squared tests were used to compare categorical variables and concussion incidence. Continuous variables were presented as the mean ± standard deviation if they were normally distributed according to the Kolmogorov-Smirnov test. Means of normally distributed continuous variables were compared using a t-test. If not normally distributed, continuous variables were presented as the median and interquartile range. Medians were compared using the Kruskal-Wallis test. Significant univariate variables were included in a multivariate logistical regression to assess for risk factors of chronic headaches. Pertinent risk factors, specifically a history of concussions, were further analyzed as a function of chronic headache and symptom burden, as reported on the PCSS survey administered at the beginning of the season. Headache burden was rated on an overall scale ranging from 0 to 54 that assessed the intensity of headache, vomiting, nausea, balance, dizziness, sensitivity to light, sensitivity to noise, numbness, and visual changes on independent scales of 0 to 6. Symptom burden included all 22 measures on the baseline PCSS survey. For the analysis, past concussions were stratified by their number and characteristics, including confusion, anterograde amnesia, retrograde amnesia, and loss of consciousness.
Next, a multivariate model accounting for variables shown to modulate chronic headaches was used to compare the incidence of future concussion among those reporting a history chronic headache and those reporting no such history. Future concussion incidence was also analyzed as a function of initial headache burden. Separate multivariable logistic regressions were then conducted to assess for the role of chronic headaches on concussion severity and recovery as measured by deviations from baseline to PI and then to FU in the five composite scores. In the multivariate model assessing concussion recovery loss to FU, latency to FU, and deviations from baseline to PI in the five composite scores were included as confounding variables. Latency to FU was defined as the time elapsed between the PI and FU tests. A final set of multivariable linear regressions were used to assess the role of baseline demographic variables and headache burden on concussion severity and recovery.
A p-value < 0.05 was considered significant for all tests. RStudio 3.6 (R Foundation for Statistical Computing) was used for data analysis and Adobe Illustrator 27.0 (Adobe Incorporated) was used for figure creation.