Most patients with cluster headache (CH) seek medical help between attacks, and it is relatively seldom that the physician has an opportunity to witness an actual attack of headache. Results of a physical and neurologic examination are negative, with the exception of a possible partial Horner syndrome on the symptomatic side. Consequently, the diagnosis is based mainly on the history of the patient. If there is only a short history of disease, the diagnosis may be difficult, but if the patient has suffered previously from several identical periods of headache, it is easy to establish a correct diagnosis.

Some features of the pain of CH are of special diagnostic importance: (a) strict unilaterality; (b) very severe intensity; (c) orbital/supraorbital and/or temporal localization (i.e., predominantly first-division trigeminal pain); (d) prominent ipsilateral autonomic features; (e) short duration of attacks (15 to 180 minutes); (f) periods of susceptibility to attacks. Most commonly there is a characteristic periodicity to the attacks that helps to distinguish CH from other types of head pain, but as is well established about 15% of patients suffer from chronic symptoms (so-called chronic cluster headache). Sometimes a diagnosis may be obscure, especially at the beginning of the course of disease. CH commonly can be unrecognized or misdiagnosed (2,47); this most likely arises because the condition is uncommon (prevalence ˜0.1%), and most of these patients have relatively infrequent attack-susceptible periods. Some patients may at times suffer from sporadic attacks having neither typical cluster nor chronic patterns, and it can be helpful to observe the patient directly in the hospital ward. The physician can obtain valuable information of the attacks and their time pattern if the patient keeps a record during the cluster period. The intensity, duration, and time of day of the attacks should be recorded, along with the effects of any medications.

In most individuals the pain is almost invariably strictly unilateral. There have, however, been exceptional case reports of bilateral pain in CH (2,20,40,49) and, even more rare, a change in side within the same attack (2). Individuals can experience attacks on both the right and left sides, more commonly in different bouts, but also during the same bout (6,20,23,26). The pain is predominantly in the distribution of the first division of the trigeminal nerve but can radiate over a wide area that includes the cheek, jaw, upper and lower teeth, nose, ear, occiput, neck, shoulder, or whole hemicranium (2,6,11,20,23,26). During attacks, there are accompanying autonomic features: a parasympathetic discharge and lacrimation with conjunctival injection, nasal congestion, rhinorrhea, and frontal sweating, as well as a sympathetic deficit: miosis and ptosis. The partial Horner syndrome occurs during acute attacks but may persist between attacks. However, in a few patients headaches are never accompanied by these autonomic symptoms or signs (7,31). There are a number of clinical features in CH that more typically occur in migraine. Up to 50% if individuals can experience nausea, less commonly with vomiting, and a similar number can experience photophobia and phonophobia (2,20). Aura symptoms are commonly not associated with CH attacks (6), but according to a recent large study, up to 14% of individuals can experience aura symptoms associated with their acute attacks (2). Most distinctive to CH is the characteristic of restlessness or agitation (2,3,6,9,45) that is now included as diagnostic criteria in the recent revised International Headache Society (IHS) classification (17). The patient often rocks the body to and fro, or he or she wanders restlessly back and forth across the floor. A simple diagnostic test (3) is to invite patients to demonstrate how they behave during severe attacks.

Clinical examination during spontaneous or provoked attacks is an additional valuable aid to diagnosis. An attack is readily provoked by giving nitroglycerin 1 mg sublingually during an active CH period (8). Preferably, the
provocation should be carried out when the patient has not had a spontaneous attack within the previous 8 hours. The test result is accounted positive if a pain attack is induced with clinical features identical to those of a spontaneous attack. The induced attack is usually preceded by a lag of 30 to 50 minute after administration of nitroglycerin and is followed by a refractory phase lasting a few hours. The test reproducibility is good, attacks being provoked repeatedly on different occasions with an identical dose. This susceptibility decreases at either end of the bout, and an attack cannot be triggered outside of a bout (8).