This systematic review was conducted as recommended by the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines ²⁵ . The protocol of this study was registered with PROSPERO, an international database of prospectively registered systematic reviews at the University of York, on 28 ^th September 2020 ( CRD42020210332).

Articles reporting headache as a symptom of COVID-19 cases were included. COVID-19 cases should be diagnosed with RT-PCR test using either nasopharyngeal and oropharyngeal swab, bronchoalveolar lavage or cerebrospinal fluid (CSF). All cross-sectional and cohort studies that included COVID-19 cases randomly selected from the population were considered eligible while case reports and case series including all editorials, reviews, and commentaries were excluded. Case-control studies with pre-allocated number of patients with headache and non-headache were excluded. Studies that were conducted in specific populations only such as in pregnancy, children, cancer patients and other groups were excluded. Only articles written in English were included in this study.

To identify potential articles for analysis, systematic searches were conducted using three bibliographical databases ( PubMed, Scopus, and Web of Science as of September 2 ^nd, 2020). The search criteria were as follows. Pubmed ([Title] "SARS-CoV-2" OR "COVID-19" OR "Wuhan coronavirus" OR "Wuhan virus" OR "novel coronavirus" OR "nCoV" OR "severe acute respiratory syndrome coronavirus 2" OR "coronavirus disease 2019 virus" OR "2019-nCoV" OR "2019 novel coronavirus" OR "severe acute respiratory syndrome coronavirus 2" OR "coronavirus" OR "coronaviruses" OR "SARS 2" OR "2019-nCoV acute respiratory disease" OR "novel coronavirus pneumonia"OR "COVID") AND ([All] “Headache”). Scopus ([Title] "SARS-CoV-2" OR "COVID-19" OR "Wuhan coronavirus" OR "Wuhan virus" OR "novel coronavirus" OR "nCoV" OR "severe acute respiratory syndrome coronavirus 2" OR "coronavirus disease 2019 virus" OR "2019-nCoV" OR "2019 novel coronavirus" OR "severe acute respiratory syndrome coronavirus 2" OR "coronavirus" OR "coronaviruses" OR "SARS 2" OR "2019-nCoV acute respiratory disease" OR "novel coronavirus pneumonia"OR "COVID") AND ([All] “Headache”). Web of Science ([Title] "SARS-CoV-2" OR "COVID-19" OR "Wuhan coronavirus" OR "Wuhan virus" OR "novel coronavirus" OR "nCoV" OR "severe acute respiratory syndrome coronavirus 2" OR "coronavirus disease 2019 virus" OR "2019-nCoV" OR "2019 novel coronavirus" OR "severe acute respiratory syndrome coronavirus 2" OR "coronavirus" OR "coronaviruses" OR "SARS 2" OR "2019-nCoV acute respiratory disease" OR "novel coronavirus pneumonia"OR "COVID") AND ([All] “Headache”). English as language limitation was imposed in the searches. Only peer-reviewed articles were included. Data were extracted both from the articles and the supplementary materials. Reference lists from the eligible articles were retrieved for further relevant studies.

All titles and abstracts of identified articles were imported into the EndNote X9 (Thompson Reuters, Philadelphia, PA, USA) and duplicate records between databases were removed. Retrieved articles were initially screened based on title and abstract to identify possible eligible studies. The full texts of potentially eligible articles were then reviewed. The screening and review processes were conducted by two authors (MF and JKF). After reviewing the full texts, the eligibility of each study was decided.

The following data were extracted from eligible articles: study characteristics (author, title, journal, study site and study design), headache characteristics (number of patients with headache, type of headache, localization, and severity), COVID-19 characteristics (number of patients with COVID-19, severity, and outcome).

This study received no external funding.

The primary outcomes of this systematic review were: a) the prevalence of headache in COVID-19 cases; and b) the association between headache and COVID-19 cases compared to other viral infections.

The cumulative prevalence rate of headache was calculated for COVID-19 cases in the general population. The prevalence was calculated as the number of COVID-19 cases with headache divided by the total number of COVID-19 cases with and without headache, expressed as a percentage (%). Pooled odd ratios (OR) and 95% confidence intervals (95% CI) were calculated to assess the association of headache and COVID-19 compared to non-COVID-19 cases.

To reduce sample selection bias, a critical assessment was specifically conducted in terms of setting of study and diagnosis of COVID-19. The quality of eligible studies was assessed using critical appraisals based on the Newcastle-Ottawa scale (NOS) ²⁶ .

The association between headache and the presence of COVID-19 was assessed by the calculation of a pooled OR and 95%CI using the Z test (p<0.05 was considered statistically significant). Prior to analysis, gathered data from studies were evaluated for heterogeneity and potential publication bias. Heterogeneity among studies was assessed using the Q test. Initial analysis found that the data had heterogeneity (p<0.10) and therefore a random effect model was employed. Egger’s test and a funnel plot were used to assess the reporting or publication bias (p<0.05 was considered having potential for publication bias). The data were analyzed using Review Manager version 5.3 ²⁷ . The cumulative pooled OR and 95%CI was presented in a forest plot.

The literature searches yielded 732 articles, of which 229 were excluded as duplicates between databases. Following a screening process of the titles and abstracts of the remaining 503 articles, an additional 253 articles were excluded due to irrelevant studies leaving 250 articles ( Figure 1). The full texts of the remaining 250 articles were retrieved and screened for eligibility. This process excluded additional 49 articles that were not eligible as they did not fulfill the inclusion criteria. A full assessment was conducted on 201 articles.

To calculate the prevalence of headache in COVID-19, full-text assessment resulted in the exclusion of 123 articles for the following reasons: case control studies (n=2), case report studies (n=69), case series (n=29), duplicated dataset (n=1), and conducted in specific population only (n=23). The targeted population studies were conducted among healthcare workers ^{28– 31} , diabetic patients ³² , pediatrics ³³ , pregnant women ³⁴ , cancer patients ³⁵ , children and young adults ^{36– 40} , patient undergone surgery ⁴¹ , critical patients ⁴² , women undergone delivery ⁴³ , patients with mild-moderate COVID-19 ^{44– 46} , patients with gastrointestinal symptoms ⁴⁷ , patients with severe headache only ⁴⁸ , and patients with anosmia only ⁴⁹ . In total, 78 studies were included to calculate the prevalence of headache in COVID-19 and all studies were published in 2020. The studies were conducted in Brazil ⁵⁰ , China ^{13, 51– 96} , Egypt ⁹⁷ , France ^{98– 101} , Germany ¹⁰² , India ^{103, 104} , Italy ^{105– 111} , Japan ^{112, 113} , Jordan ¹¹⁴ , Somalia ¹¹⁵ , South Korea ^{116, 117} , Spain ^{118, 119} , Turkey ^{120, 121} , and the US ^{122– 126} . Two studies were cross-sectional ^{119, 122} , five were prospective cohort studies ^{65, 90, 98, 110, 121} and the remaining 71 studies were retrospective studies.

To calculate the association between headache and COVID-19, the full-text assessment yielded 16 eligible studies. The rest of the references had been excluded for these reasons: (a) the studies were case reports or case-series (n=98); (b) the full-text did not include data of outcome of interest (n=84); and (c) low quality of study (n=3) ( Figure 1). The included studies were conducted in a wide ranges of regions: Australia ¹²⁷ , Belgium ²⁹ , Brazil ⁵⁰ , China ⁷⁰ , France ⁹⁹ , Hongkong ¹²⁸ , Israel ¹²⁹ , Italy ¹³⁰ , Germany ¹³¹ , Netherlands ³¹ , Turkey ¹²⁰ , and the US ^{28, 105, 122, 132, 133} . Out of the studies, ten were case-control ^{28, 29, 31, 99, 105, 120, 128, 130, 132} , four were cross-sectional ^{50, 70, 122, 131, 133} , and two were prospective cohort studies ^{127, 129} .

Our systematic review included 78 studies consisting of 104,751 COVID-19 patients and headache was reported in 26,464 patients with a cumulative prevalence of 25.26%. The list of the studies and the prevalence of headache of each study is presented in Table 1. One study which included 51 patients described the specific location of headache: 1.96% (1/51) was a temporal headache, 35.29% (18/51) was a frontal headache, 23.52% (12/51) was a retro-orbital headache, and 39.21% (20/51) was a diffuse headache ⁹⁹ . Another study which involved 46 patients reported that 40 (86%) had tension-type pain and 6 (14%) had migraine-like headache ¹⁰⁷ . Data from 18 studies indicated that 72.17% (236/327) of headaches were reported in mild-moderate COVID-19 cases ^{51, 53, 55– 58, 62, 63, 67, 76, 88, 92– 97, 117} . The prevalence of headache in severe COVID-19 cases from 15 studies was 27.83% (86/309) ^{53, 55, 57, 58, 62, 63, 73, 76, 88, 92, 93, 95– 97, 117} .

A total of 16 studies, consisting of 5,407 COVID-19 cases in adults and 94,818 adults with non-COVID-19 infections (mostly COVID-19-like respiratory viral infections), were analyzed to determine the association between headache and COVID-19. Of these studies, an association between headache and the occurrence of COVID-19 was observed in 9 studies ^{28, 29, 31, 105, 122, 127, 129, 130, 132} while seven studies reported no association ^{50, 70, 99, 120, 131, 133, 134} ( Table 2). Our cumulative calculation revealed that headache was found to be 1.7-fold more prevalent in patients with COVID-19 compared to those with non-COVID-19 respiratory viral infections, OR: 1.73; 95% CI: 1.94, 2.51 with p=0.04. The correlation between headache and COVID-19 is presented in Figure 2.

As a non-specific symptom, headache might present not only in COVID-19 cases but also in other viral diseases, therefore, headache might not raise suspicion of SARS-CoV-2 infection ^{22, 23} . However, a study described that headache is one of the main neurological symptoms of coronavirus infection including SARS-CoV-2 ¹³⁵ . The global prevalence of headache in our systematic review is more than 25% out of 104,751 COVID-19 cases. This result was almost double that of the previously reported prevalence from studies in China early in the pandemic that ranged from 6.5–13.1% ^{53, 88} . This suggests that headache is prevalent in SARS-CoV-2 infection and therefore could potentially be used as one of the indicators to diagnose COVID-19 cases.

Our study also highlights that headache was significantly more prevalent in COVID-19 patients, 2.2-fold, than suspected non-COVID-19 viral infection. A study found that only around 11% of MERS patients reported they suffered from headaches ¹³⁶ . A study in COVID-19 patients with pre-existing primary headache disorders revealed that the headache during COVID-19 had an unusual presentation with 42% (44/104) reporting a recent onset of headaches, 49% (51/104) had a change in headache pattern, and 39% (39/104) reported the worst headache they had ever had ¹¹⁹ . These results suggest that new onset headache and changes of headache pattern should be carefully explored as this might be able to differentiate patients with COVID-19 from those without.

We explored the available literature to broaden our knowledge of the pathogenesis of headache in COVID-19. In general, three main primary headaches are observed in COVID-19 patients i.e. migraine, cluster headache and tension-type headache ^{137– 140} . No fixed mechanisms have been reported on how these headaches emerge in COVID-19 patients. However, it has been proposed that the activation of trigeminal nerve ending in the periphery followed by the sensitization of various sites in the brain is one of the main pathomechanism of headache in these patients ^{137, 138} .

A headache attack is initiated by the release of several vasoactive neuropeptides such as glutamate, calcitonin gene-related peptide (CGRP), substance P and pituitary adenylate cyclase-activating polypeptide (PACAP), from nociceptive sensory fibers (especially nociceptive C-fibers and Aδ-fibers) innervating blood vessels located in the meninges and other cranial structures leading to vasodilation, degranulation of mast cells and plasma protein extravasation in those vascular structures ^{141– 143} . The release of those peptides from the fibers could be due to either electrical, chemical or mechanical induction emerged from three branches of the trigeminal nerve i.e. ophthalmic, maxillary and mandibular branches ¹⁴³ . However, because of its wider area of innervation in the meninges and cranial blood vessels, ophthalmic branch seems to play more of a role in stimulating nociceptive processes in meningeal structures than the other two branches ^{143, 144} .

Next, any physiological events, such as vasodilation, that have occurred in the meningeal and large cerebral blood vessels will become a stimulus which is sent to the trigeminal ganglion (TG) where other nociceptive information from other afferent trigeminal branches are also converging ¹⁴¹ . Although cerebral and meningeal vasodilation is not the sole cause of headache ¹⁴⁵ , most studies have agreed on the critical role of blood vessel dilation in the emergence of headaches.

From the TG, the stimulus is projected to an area in the brainstem called the trigeminocervical complex (TCC) via first-order neurons ¹⁴³ . These transmissions are then projected to the diencephalon structures, including the thalamus and hypothalamus, via the second-order neurons ^{143, 146} . The third-order neurons are subsequently responsible for transmitting the information from diencephalic systems to various cortical areas associated with motoric, somatosensory, auditory, retrosplenial and visual functions ^{141, 143} , leading to the manifestation of headache pain and other related symptoms ( Figure 3).

During these transmission processes, the release of neuropeptides, especially CGRP, is limited only in the meninges and in the central terminals of trigeminal afferents ¹⁴⁷ . When the transmission reaches TCC structures, CGRP and substance P may act to induce the release of glutamate and reduce gamma aminobutyric acid (GABA) production ^{147, 148} . It has been proposed that during a headache attack, the level of glutamate in the TCC increases, while GABA release is decreased ¹⁴⁸ . This condition could result in the increase of nociceptive neurons excitability ¹⁴⁹ . Moreover, low level of serotonin in trigeminal nerve might also be involved in migraine pathophysiology as the release of this neurotransmitter has been linked to the inhibition of CGRP in trigeminal nerves ^{150, 151} .

Several mechanisms have been postulated to explain how trigeminovascular system is activated in COVID-19. Firstly, direct invasion of the virus may activate the peripheral trigeminal system ¹³⁷ . This theory is hypothesized according to a study, confirming that trigeminal ganglia possess an angiotensinergic activity ¹⁵² . Thus, the viral attack would hypothetically disturb the activity of the renin-angiotensin-aldosterone system (RAAS), which may increase the level of CGRP ¹⁵³ . Although the invasion of SARS-CoV-2 into the olfactory nerve ending seems to be the main route ^{154– 156} , the action of the virus on the trigeminal nerve must not be overlooked, as suggested by Perlman et al. (1989). They demonstrated that the trigeminal nerve, in addition to the olfactory nerve, was a route used by neurotropic murine coronavirus to invade the central nervous system (CNS) ¹⁵⁷ . The hypothesis of trigeminal nerve attack by SARS-CoV-2 is also supported by the fact that olfactory mucosa is innervated by the trigeminal nerve ^{158, 159} suggesting the invasion of the olfactory mucosa by SARS-CoV-2 may also induce trigeminal nerve injury.

Following entry into the trigeminal nerve, SARS-CoV-2 may hijack the transneuronal transport system to direct the virus to enter the nucleus via a retrograde axonal transport mechanism. This transport occurs by the involvement of cytoskeletal motor proteins called dynein supported by cofactor dynactin that function to move substances, such as endosomes and vesicles, including hijacking viruses, on microtubule towards the cell body ¹⁶⁰ . Once the virus gains access to the nucleus in the cell body through the microtubule-organizing center (MTOC), viral replication is initiated ¹⁶¹ . Finally, viral progenies may spread to other areas of the body, including the CNS, via anterograde axonal transport assisted by the kinesin motor protein family ¹⁶⁰ .

A study suggested a transneuronal transport system used by coronavirus after investigating the neuroinvasiveness of HCoV-OC43 in mice ¹⁶² . Furthermore, the movement of SARS-CoV-2 via retrograde axonal transport is also hypothesized as it has been reported that the envelope protein of SARS-CoV could subvert dynein function either directly or indirectly ¹⁶³ . The role of dynein in the retrograde axonal movement of several viruses, such as herpesviruses, West Nile virus, rabies, and influenza virus, upon their penetration in the neuronal plasma membrane has also been reported ^{164– 169} .

Secondly, SARS-CoV-2 may also invade the trigeminal nerve by indirect mechanisms. Cytokine storm and vasculopathy mechanisms are also proposed to explain the activation of trigeminal nerve upon SARS-CoV-2 infection ¹³⁷ . Cytokine storm has attracted substantial interest from researchers and clinicians as this unwanted condition is strongly suggested to be related to the increased mortality in SARS-CoV-2-infected patients ^{170, 171} . It is hypothesized that the headache suffered by COVID-19 patients at the later stage of this infection is induced by the cytokine storm ^{139, 172, 173} . This notion has been supported by the fact that proinflammatory cytokines, such as IL-1, IL-6 and TNF-α, have been linked to the activation of the trigeminovascular system, which is responsible for the emergence and development of headache through the modulation of CGRP ^{174– 177} .

Moreover, the presence of angiotensin-converting enzyme 2 (ACE2) receptor on the endothelial cells makes the blood vessels vulnerable to invasion by SARS-CoV-2 ^{178, 179} . It is known that ACE2 is associated with several protective mechanisms within the body, such as vasodilation ¹⁸⁰ and antinociception ¹⁸¹ . ACE2 also diminishes excessive free radical production which prevents oxidative stress ¹⁸² . The utilization of this receptor by the virus may decrease its activities, leading to the disturbance of vascular function. The perivascular trigeminal nerve may in turn be affected resulting in the COVID-19-related headache ¹³⁷ . More studies are required to improve our understanding on the role of the ACE2 receptor in headache pathophysiology.

Another hypothesis by which SARS-CoV-2 could induce headache is offered by Abboud et al. (2020). They proposed that gas exchange disturbance in alveolar tissues triggered by the viruses would induce hypoxia, which in turn leads to ischemia ^{170, 183} . Ischemia itself has been known to have a strong relation with headache incidents ¹⁸⁴ that could be induced by exaggerating the production of free radicals.

In regards to the headache characteristics presented by COVID-19 patients compared to the headache induced by other viral infections, no apparent differences could be observed. Headache in COVID-19 could be worsened by physical or head movement, felt in either the entire head (holocranial) or unilaterally (hemicranial) and and the pain is typically pressing or tightening ¹³⁹ . It is hypothesized that headaches occurring in COVID-19 patients might be the result of the same mechanisms as observed in influenza A and influenza B infections, which could be related to the activity of cytokines ^{173, 185– 187} . A recent report on dengue-related headache suggested that the headache could be pulsating and either affect the entire brain, only frontal, or orbital area, which may resemble primary headaches reported in COVID-19 patients ^{139, 172, 188} . Therefore, although we found that headache is more frequent in COVID-19 patients than those of non-COVID-19 patients, diagnosis of COVID-19 should not be based on the presence of a headache.

In conclusion, headache is a common symptom in COVID-19 cases. Some mechanisms have been proposed as to the mechanism for headaches in COVID-19 such as the activation of the trigeminovascular system by either direct action of the virus or indirect mechanisms induced by cytokine storm, vasculopathy, or ischemia induced by gas exchange disturbance in COVID-19 patients. Extensive efforts must be carried out to provide definitive answers about COVID-19-related headaches. Detailed investigations on the mechanisms by which SARS-CoV-2 attacks the CNS and thus generates headaches are important to improve our understanding on the pathophysiology of COVID-19, and therefore influences possible pharmacological intervention decisions.

All data underlying the results are available as part of the article and no additional source data are required.

Figshare: PRISMA checklist for ‘Global prevalence and pathogenesis of headache in COVID-19: A systematic review and meta-analysis’, https://doi.org/10.6084/m9.figshare.13166783.v1 ¹⁸⁹

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).