The prevalence of primary headache disorders in children and adolescents in Zambia: a schools-based study

This second enquiry of its type to be reported from SSA found headache disorders to be common among children and adolescents, as did the first in Ethiopia [6] and others using similar methodology in Turkey [17], Lithuania [18], Austria [25] and Mongolia [19]. In this study, according to symptoms reported in response to the questionnaire used in all these studies [5], almost all of the young participants had experience of headache (lifetime prevalence 97.5%), and most (85.8%) had an active headache disorder (at least one episode of headache in the preceding year). These levels exceeded those found in all other studies [6, 17,18,19, 25]. Again according to symptoms reported, migraine was the most reported type. To this extent, this study agreed with the others (Table 6), but the estimated migraine prevalence of 53.2% (including 35.7% probable) rendered this study an outlier, raising questions as to validity. We return to this below. In other findings, UdH (14.8%) was second most prevalent, and within the range of the other studies (Table 6); TTH (12.1%) was less common than in other studies; headache on ≥ 15 days/month was evident in 3.3%, of whom 0.9% also reported overuse of medication (pMOH). Very few headaches (2.3%) remained unclassified.

Table 6 1-year prevalences (% [95% CI]) of all headache, migraine, tension-type headache and undifferentiated headache in this and other studies using the same questionnaire [5]

Before commenting further on these findings, we should note the study’s several limitations, since the findings need to be considered in their light. There were logistic and practical difficulties, not least the high numbers per class in some schools, and the poor literacy requiring questions to be read and explained to many children. In addition, the cholera outbreak mid-study was exceedingly disruptive [20].

On a general level, the principal limitation was uncertainty over the quality of responses. The same problem was noted in Ethiopia [6], and may to a degree be unavoidable. Mediated enquiry may mitigate this somewhat but, as we found here, far from entirely. As also noted in the Ethiopia study [6], epidemiological diagnosis of headache type in children and adolescents appears to be inherently imprecise. This is evidenced also by the many past surveys that have simply left large numbers of reported headaches undiagnosed (these are discussed in [17]). The questionnaire could not be validated in the local languages used in the study for the same reason as in the adult study [7] and in Ethiopia [6]: no headache experts in the country with availability to do this. It may be that some questions need reformulation, particularly those structured as leading questions enquiring into associated symptoms [5], although these have not proved problematic in Turkey [5, 17], Austria [5, 25], Lithuania [18] or Mongolia [19]. On the other hand, in these age groups, many headaches are evolving from an undifferentiated form [17], so diagnostic ambiguities are not unexpected. For this reason, extending the enquiry to parents or carers would carry no guarantee of greater certainty – and responses would not be better informed.

Other study limitations were on a specific level. The sampling process failed to equalise numbers across the age groups – a consequence of the practical difficulties in engaging with younger pupils described earlier. Additionally, schools-based sampling in Zambia has known inbuilt bias against female adolescents [9], for which statistical correction can be applied, but also likely (but unknown) degrees of bias against potential participants from low-income and rural areas. Zambia is not a low-income country by World Bank classification [26], but the reality on the ground in the more rural areas is of poverty. On the other hand, Zambia is relatively highly urbanised (44% against the SSA average of 41% [27]).

In the light of these limitations, we consider – and question – the high apparent prevalence of migraine (53.2%). Among the studies so far conducted within this global programme, this one is a clear outlier in this regard, surpassing even Ethiopia [6] (Table 6). Table 3 reveals that, while headache duration was ≤ 2 h in nearly three quarters, 60–76% of participants with headache reported each of three characteristics suggestive of migraine [21]. Only unilaterality, recognised as less common in young people, was reported by a minority (31%). Among accompanying symptoms, almost 90% reported phonophobia as an accompanying symptom. In Ethiopia, almost 80% did so [6]. These high proportions render this supposedly migraine-specific symptom unhelpful in differential diagnosis. Phonophobia makes no contribution to diagnosis on its own but, with 90% reporting it, the presence of photophobia (47.5%, itself a high proportion) would almost certainly mean fulfilment of one key criterion for migraine [21]. Against this, only 28.6% met the duration criterion for migraine (minimum 2 h [21]), so that most migraine diagnoses could only be probable, as two thirds were. It is, in fact, the finding of 35.7% probable migraine that is so unlikely. Over one third (36.6%) of these were reported as lasting ≤ 1 h, and the diagnosis of probable migraine rather than UdH rested entirely, therefore, on reported intensity. If these headaches were recategorized as UdH, the estimated prevalence of migraine would be reduced to 39.0%, much in line with Ethiopia’s 38.6% (Table 6), but this finding, too, was questioned as improbably high [6]. As an estimate of migraine prevalence, we can be confident only of the 17.5% definite; this estimate, however, is below all others in Table 6, and the reality is almost certainly higher.

Females reported slightly but significantly more headache overall, while differences in headache types were small (just significant for migraine [AOR: 1.2; p < 0.05; Table 4]). Age-dependent differences were seen only in TTH (AOR: 1.4) and all headache on ≥ 15 days/month (AOR: 1.8), both more prevalent (p < 0.05) in adolescents than children. UdH did not decline in prevalence with age (Table 4); neither did it decline as a proportion of all headache (children 16.2%; adolescents 17.3%). Since UdH is understood to represent expressions of migraine or TTH by the immature brain [17], an inverse relationship with differentiated headache types is expected as the latter develop with increasing age. This was seen in Turkey [17], Lithuania [18], Austria [25] and Mongolia [19], but not Ethiopia [6] – of relevance since Ethiopia is also in SSA. The explanation in both countries probably lies in the diagnostic uncertainty between migraine and UdH when headache is short-lasting and diagnosis depends, essentially, on reported intensity – subjectively, and on a rather insensitive scale of 1–3.

There was a high prevalence of headache on ≥ 15 days/month, reported by one participant in every 30 (3.3%). This was a much higher proportion than in Ethiopia (1.3% [6]), but it was given credence by the finding that 59% of these reported HY (Table 5). In the adult study in Zambia, the prevalence of headache on ≥ 15 days/month was egregiously high (11.5%), with pMOH accounting for most of it (7.1%) [7]. This was also in contrast to Ethiopia, where the adult prevalence of headache on ≥ 15 days/month was only 3.2% and of pMOH 0.7% [28]. As can be seen, pMOH accounted for most of the difference, and, in the more urbanised Zambia [27], pMOH was very largely an urban problem [7]. The messages from the present study – of public-health importance – are not only that headache on ≥ 15 days/month is already common in young people but also that its prevalence is detectably increasing with age (2.6% observed in children, 4.4% in adolescents; Table 2).

While HY was reported by well over half (59%) of those with headache on ≥ 15 days/month, this was supported by estimated 1-day prevalence based on reported mean frequency, which, in the preceding 28 days, was actually somewhat higher (70%). For the episodic headaches, however, reported headache days in the preceding one and four weeks always led to predicted proportions with HY that were lower than reported proportions (Table 5). The differences were considerably greater when predictions were based on four weeks rather than one. The same phenomenon was seen in Ethiopia [6] and Mongolia [19], where it was suggested that, while HY reporting should be free from it, recall error might increasingly influence retrospective estimates over longer recall periods. But, overall, 22.2% (464/2,089) of these young people reported HY, reflecting a presumed 1-day prevalence that is concerningly high. It is, nonetheless, a lower proportion than the 26.9% in Ethiopia [6].

Finally, we acknowledge that malaria is very common in Zambia (although much less so in Lusaka) [29]. Its symptoms conspicuously include headache, to the extent that malaria has been (and in some places still is) the default diagnosis of headache. The enquiry with respect to headaches in the preceding year established that most were episodic, recurrent and short-lasting. This would be an unlikely presentation of malaria, or of other prevalent infectious diseases of which headache is commonly symptomatic. We cannot say this of the small proportion (< 3%) reporting headache on ≥ 15 days/month, or of the 2.3% whose headaches remained unclassified.

The strengths of this study lie in the tested and validated methodology [5] and adequate sample size [10]. Limitations are identified and discussed above.

留言 (0)

沒有登入
gif