Having shown headache to be highly prevalent among adults in Cameroon [1], here, using data collected contemporaneously from the same participants, we report high levels of burden attributed to it. On average, those reporting any headache in the last year (by definition, an active headache disorder [12]) spend 9.8% of their time with headache (pTIS) of moderate intensity (2.2 on the scale of 0–3). pTIS is, of course, substantially higher among the 13.1% with H15 + (pMOH 52.4%; other H15 + 31.0%) than the 18.1% with migraine (4.6%) or the 44.8% with TTH (2.3%). At population level, a total of 6.1–7.4% of all time is spent having headache, and this is, on average, associated with lost time from income-generating work of 2.5 days/3 months. As an economic indicator of burden, likely to be reflected in national gross domestic product, this is very high indeed (3.8%, assuming a 5-day working week).

Higher productivity losses were seen among those with H15 + than those with migraine or TTH, reflecting the differences in pTIS. Especially, pMOH was associated with estimated losses of 10.8 days/3 months from paid work and 9.8 days/3 months from household work. With its very high prevalence of 6.5% [1], this meant that pMOH contributed more than one quarter (0.7 days/3 months) to all headache-attributed lost income-generating work time (2.5 days/3 months). The importance of this, to health and economic policies, lies in the fact that MOH is an avoidable illness, although avoidance requires public education.

But the impact of TTH on lost productivity is noteworthy: two-fold that of migraine (1.0 vs 0.5 work days, and 0.9 vs 0.4 household days). This is because of its high prevalence: pTIS for TTH was lower than for migraine (owing to shorter headache duration), while estimated lost health from TTH was substantially lower, reflecting both the lower pTIS and the much lower DW (0.037 compared with 0.441 [3, 13]). (Here it should be noted again that DWs, as used by the Global Burden of Disease study [3], are a broad measure of lost health rather than disability [15, 16]). The mean reported headache intensity was indeed higher in migraine than in TTH (2.6 vs 2.2 on the scale 1–3, which may indicate severe vs moderate), but by far less than the difference in DWs. While lost productivity might be expected to correlate with symptom burden (more severe headache leading to greater lost productivity), our earlier study has shown headache frequency to be the main driver of lost productivity [17], intensity less so and duration having virtually no impact. Since mean headache frequency was reportedly the same in TTH and migraine (2.8 days/month), the two-fold higher lost worktime for TTH can be well explained by its more than two-fold prevalence (44.8% vs 18.1%) – but this calls for a rethinking of TTH as a generally non-disabling headache [18]. It is possible that, among those diagnosed as TTH, some in fact had migraine. This would have increased the prevalence estimate for TTH while reducing that for migraine, and elevated the burden attributed to TTH. However, the prevalence estimates were very close to those in nearby Benin (TTH 43.1%, migraine 21.2% [6]), suggesting any such error was small. The key message here is that, in Cameroon, although TTH may be less burdensome at individual level than other headache types, its impact at population level is not to be overlooked in health and economic policies.

We saw no significant gender-related differences in frequency, duration, pTIS or impaired participation in paid or household work or leisure activities with regard to H15 + . Migraine, on the other hand, was associated with higher headache frequency, and consequently higher pTIS, among females than among males, and, probably consequentially (wholly or partly), with greater productivity losses from both paid and household work among females than among males. A similar pattern regarding pTIS and lost productivity was seen for TTH, even though mean headache frequency and duration were similar in the two genders (a reflection of skewed data).

With the inclusion of enquiry into HY, and factoring in its prevalence, we had two sets of data for estimating symptom burden and impaired participation. Although HY data were inevitably based on a lower N, they were presumably free from recall error. We have already demonstrated that predicted 1-day headache prevalence (estimates made from 1-year prevalence and reported frequency) tends to be lower than observed 1-day prevalence (ie, of HY) [1]. This finding suggests recall underestimates frequency. Here, since we had information on duration of HY, we were also able to calculate pTIS solely from HY data, but, somewhat surprisingly, this gave a slightly lower estimate than the calculation based on headache frequency in days/month and usual headache duration (6.1% vs 7.4%). Recall perhaps overestimates duration. Nevertheless, these two independent calculations corroborated each other in showing that a very substantial proportion of all time in Cameroon is spent having headache.

In contrast, direct comparison between impaired participation calculated from HALT and HY data is not straightforward. Importantly, our method of calculating impaired participation from HY (counting less than half as nothing done, and more than half as everything) has not been validated in the same way as for HALT [14]. It does not make the same distinction between absenteeism from and reduced productivity with headache while at work, but, rather, accepts respondents’ subjective estimates of their actual overall activity yesterday in relation to their intended activity. Furthermore, it does not differentiate between different domains of participation. Still, the estimated 6.9% impairment in participation derived from HY is very high, and underpins our finding (based on HALT) that headache leads to very substantial losses in productivity.

High levels of headache-attributed impaired participation were also reflected in the extremely high proportion of Cameroon’s adult population believed to be in need of headache-care (37.0% when adjusted for age and gender). H15 + (13.1%) is the leading call on headache care, but TTH follows closely: with an estimated 8.2% of the population losing ≥ 3 work and/or household days/3 months from TTH, age- and gender-corrected estimates showed 12.8% of the population to be in need of headache care for this very neglected disorder. A somewhat lower but far from insubstantial 11.3% require headache care for migraine.

The data on QoL deserve comment. WHOQoL-8 is a generally insensitive measure, without intuitively meaningful units, but it showed significantly lower QoL among those with H15 + than among those with migraine, TTH or no headache. High individual symptom burden in H15 + is the most likely explanation, although, despite the symptom burdens associated with migraine and TTH, WHOQoL scores for these did not differ from those for no headache.

WTP was descriptively highest in migraine, without statistical significance. It is a highly subjective measure, with no means of assessing veracity. The large differences between means and medians also demonstrated that the data were heavily skewed.

An account of the study strengths and limitations has been given previously [1], but is repeated here. The study used established methodology, generated a large sample representative of the country, and put quality-control measures in place. But, as in all such cross-sectional studies, there was dependence on recall, with diagnoses based solely on responses to a diagnostic question set. With a lack of resources (in particular, lack of headache specialists) to validate this question set directly in the population of interest and in the local translation, we depended on its use previously in 20 countries and almost as many languages, with direct validation in four [19,20,21,22].