We included studies of adults (>18 years) complaining of, or presenting with, PN and/or vertigo caused by confirmed central nervous system (CNS) pathology. Studies were included if the CNS pathology was confirmed either on imaging (MRI or CT), autopsy, following surgical intervention or other examinations, such as cerebrospinal fluid analysis. Non-humans studies were excluded.

We included studies that used positional maneuvers as a means of observing CPN across hospital, university, or research settings.

The initial search was for randomized controlled trials and controlled trials. However, when information from controlled trials was not available, cohort studies were eligible for inclusion. The intention was to exclude case series and case reports from the review due to the high potential for bias in these study designs. However, if there proved to be a lack of cohort studies available, published work from all study designs were accepted, including case series and case reports.

Studies were accepted for inclusion in the review if they were English, French, or German, with the exception of translations of publications.

A sensitive systematic protocol for database searching was adopted. Each topic was defined individually for the database using the “or” element. In this review, the individual topics were (central/non-benign) and (PN/vertigo). “And” was then used to connect the topic defining “or” searches to locate papers directly relating to PN or vertigo that was central or non-benign (rather than BPPV).

To ensure sensitivity of the search, “exp” (Explode) was used to search for any papers that have been assigned to the database subject heading “positional,” “vertigo,” “nystagmus,” or “pathologic,” and for any papers assigned below “positional vertigo,” “nystagmus,” or “pathologic” in the hierarchy. The “$” was applied to request the database to search for multiple possible endings to the word “position.” Positional was included in the search by means of using the “adj” (adjacent) operator. “adj3” was used to retrieve any instances of the word position/positional/positions/positioning, occurring within (three) words of vertigo, nystagmus, or vertiginous in the title or abstract of a paper.

Searches of electronic journals were conducted using the Cochrane Ear, Nose, and Throat, MEDLINE (PUBMED), the Cochrane database, and EMBASE. Google scholar and Yahoo Internet searches were also employed to identify any additional relevant material. There was no date restriction applied to the search. Searches were updated on a monthly basis between March 2014 and February 2017.

The reference lists of identified studies were scanned for further studies.

Four authors (Nora K. Macdonald, Doris-Eva Bamiou, Yougan Saman, and Amal Al-Shaikh Sulaiman) independently reviewed the full texts of the remaining 86 studies for eligibility to ascertain the final studies to be included in the review. Any disagreement was resolved by consensus or third party adjudication. A total of 28 studies were eligible for inclusion in the systematic review. The reasons for rejecting the remaining articles are demonstrated in e-Bibliography. The accepted records were grouped by study design type, of which there were two: case studies and case series.

The study selection process of this review did not identify any controlled studies relevant to the diagnosis of central positional vertigo/nystagmus. For this reason, this review included case studies and case series for consideration of their evidence. The authors recognize that such study designs, which are retrospective in nature, are inherently susceptible to the risk of bias. The case series did not typically recruit consecutive series of individuals from multiple centers, and this introduced a risk of selection bias. The studies were also at risk of detection of reporting/observer bias in which the clinical presentation of the CPN reported was not standardized. Furthermore, none of these case studies or case series that were considered relevant to the current review referred to results in a comparison group.

Despite the methodological difficulties associated with case studies/series, the authors viewed this lack of high quality evidence as justification for the review. For this reason, the review proceeded despite the inclusion of only case studies and all their associated scientific weakness.

This included the design of the study (case series or case study) as well as the number of participants and the study setting.

This included demographic details of the subject/s with reference to relevant medical history and history of the positional vertigo.

The characteristics of nystagmus upon positional testing were extracted from each patient. This description was divided into the classical features of PN, i.e., the direction of nystagmus, provoking position, duration of nystagmus, latency of onset, fatigability, and the presence of concurrent symptoms such as subjective vertigo, nausea, and/or vomiting.

The site of lesion was extracted from each subject. This information was in the form of imaging results, autopsy reports, or reports from surgical investigations.

Associated neurological signs or symptoms, or the absence there-of, were noted for each subject. This information referred to any central symptom such as gait disturbances, abnormal oculomotor function, and atypical symptoms such as unexplained weight loss, as well as reference to normal or abnormal observations upon clinical examination and testing.

Although not considered a primary outcome of this review, the respective management of each subject was extracted. This information, when present, was considered relevant in order to investigate any potential failure to respond to repositioning maneuvers, which may contribute to the diagnosis of a central, rather than peripheral, origin. In addition, reports from any surgical intervention may stand as evidence of the site of lesion.

Participant samples in the included papers involved individuals with CPN, i.e., had atypical positional or positioning nystagmus attributed to a confirmed central lesion. Subjects within each of the studies’ sample who did not have confirmed CPN were excluded. This included 88 participants in the study by Bertholon et al. (7) and 33 participants in Maire and Duvoisin (11) with peripheral PN.

Included studies ranged in (relevant) sample sizes from single case subjects (12–25) up to 14 subjects (26–29).

In total, there were 82 participants (age range 20–81 years) from 14 countries, with the majority (84%) from University settings. It was not possible to establish the mean age and the male/female ratio of the included participants due to insufficient demographic data.

Seventy-two (87.8%) study participants had central pathologies confirmed on either MRI (4, 6, 12, 13, 15–20, 22, 24, 30, 31) or CT (14, 21, 22, 31–34) imaging, and two (2.4%) upon autopsy examination (32, 34). In the remaining eight (9.8%), CNS pathologies were confirmed on imaging, but nine participants for whom the diagnostic means were not specified were excluded from the review.

Positional testing was used to provoke the CPN in all participants (see Tables S1 and S2 in Supplementary Material). In 13 (15.9%) cases, CPN was triggered by all three Dix–Hallpike (D–H), supine roll, and straight head hanging (SHH) maneuvers (4). In five (6.1%) cases, CPN was reported on both D–H and SHH (4, 16), while in one case, the CPN was triggered by D–H and supine with head to the left (18).

In 33 (40.2%) cases, D–H alone was used to trigger CPN (7, 11, 12, 14, 17, 19, 25, 27, 31, 34). CPN was triggered only upon lying supine with either ear down in 26 (31.7%) cases (6, 7, 13, 15, 20–24, 26, 28, 32–35) and only upon SHH in 3 (3.7%) (6, 31, 33). In one case, CPN was triggered by SHH and horizontal head rotation with either ear down (26).

A repeat of the positional test (to observe a fatigue effect) was reported in 28 (34.1%) subjects (6, 12, 13, 15, 17, 19, 22, 31, 34).

Additional evaluation often included a clinical or objective examination of oculomotor (85.4%) (4, 6, 7, 11–19, 22, 24–26, 28–30, 32, 34, 35) and gait (41.5%) function (6, 14–17, 19, 22, 26, 27, 30, 31, 34).

The means of assessing the observed nystagmus upon positional testing was poorly reported. PN was quantified in 51 (62.1%) participants, 35 (42.7%) with VNG (4, 15, 20–26, 35), 15 (18.3%) with electronystagmography (11, 14, 18, 33), and 1 (1.2%) using three-dimensional scleral induction coil (6). It is uncertain how the nystagmus was recorded in the remaining 31 participants (37.8%), presumably from direct observation during clinical examination (6, 7, 12, 13, 16, 17, 19, 30–34).

The study set included 11 case series (6, 7, 11, 26, 30–36) and 17 case reports (12–25, 27–29). This review did not identify any controlled studies relevant to the diagnosis of central positional vertigo/nystagmus.

The direction of nystagmus and provoking positional test were sufficiently reported in 70 (85.4%) patients only, and atypical direction of the nystagmus, on the basis of that predicted for the canal stimulated during positional testing, was a prominent feature.

The duration of CPN upon D–H was reported in 36 (43.9%) patients, the majority of which was paroxysmal (n = 21, 58.3%), lasting less than 16 s in 19 (52.8%) (4, 14, 16, 19, 27, 34) and “transient” in 2 (5.6%) (31, 34). In the remaining 15 (41.7%), the CPN upon D–H lasted at least a minute (11, 17), or was referred to as “persistent” (12, 18, 30).

Of the 20 subjects in which duration was reported upon SHH, 17 (85%) demonstrated transient CPN (of <16 s) (4, 6, 16). CPN persisted for over 1 min upon SHH in the remaining 3 (15%) subjects (4, 33).

In contrast, of the 24 subjects who reported the duration of CPN upon supine roll testing, the majority (n = 15, 62.5%) was persistent, lasting at least a minute (4, 6, 13, 18, 22, 24, 32, 33), while the remaining 9 (37.5%) was short-lived (less than 1 min) (4, 6, 15, 20).

Of the 35 patients with pDBN, 25 (71.4%) reported its duration. pDBN was short-lived (<17 s) in 21 (60%) cases (6, 14, 16, 27) and persisted for at least 1 min in only 4 (11.4%) subjects.

In total, across all positional tests, CPN persisted for at least 1 min in 33 out of 81 (40.7%) in occasions it was reported.

Latency of onset of PN upon assuming the provoking position was reported in 38 (46%) patients. Nystagmus “without latency” upon positioning was observed in 30 out of 38 (78.9%) patients (4, 6, 15–17, 19, 21, 22, 25, 30, 32, 34). Six (15.8%) reported CPN with a latency lasting either a “few” seconds, <2 s, or with a “short” latency upon positioning (6, 13, 14, 30, 31, 34). The remaining two cases (5.3%) reported CPN with an onset >3 s (3–5 and >10 s) upon positioning (6, 18).

pDBN developed after a short latency period (<2 s) in all six (100%) cases in which this feature was reported (6, 14, 16, 30, 31).

The nystagmus response upon repeated testing was reported in 28 (34.1%) participants. The CPN fatigued with repeated maneuvers only in 5 (17.9%) of these cases (12, 13, 31, 34), but did not fatigue in the remaining 23 (82.1%) (6, 15–20, 22, 34).

Nystagmus with no remission was demonstrated by Imai and colleagues (20) who reported a patient’s pHN beating away from the lowermost ear on left and right supine that continued without remission for 1,600 days. No patient (n = 16) (13, 19, 22, 24–26, 35) responded to repositioning maneuvers.

The effect of fixation on CPN was only reported in four studies. Maire and Duvoisin (11) found that, in a sample of 43 patients with static PN, the predictive value of the ocular fixation test was 94% (n = 35) for peripheral lesions and a 100% (n = 8) for central disorders, with the latter being associated with reduced optic fixation index. Buttner et al. (1, 6) reported poor vestibular ocular reflex (VOR) suppression (6), while Williams et al. (25) reported normal VOR suppression (25). Furthermore, Barber (14) reported pDBN, which was indeed enhanced with fixation, and Cobb and Friedman (18) reported nystagmus, which disappeared when fixation was removed.

It is noteworthy that only 2 (2.4%) patients within the data set reported the presence of spontaneous nystagmus (12, 25). This minimized the potential influence of the well-known phenomenon that a discrete spontaneous downbeat nystagmus, which may not be visible under Frenzel glasses, becomes only evident by positioning the patient.

While it is proposed that a central origin has to be assumed for pure upbeat, downbeat, and torsional nystagmus (1), we found that in about a one-third of cases (29.9%), CPN was a mixture of horizontal, torsional, and vertical components in variable combinations depending on the positioning maneuver performed. Furthermore, the direction of CPN is not always “atypical” for the plane of the canal being stimulated, since downbeat nystagmus was the most prominent direction during SHH (in 77.3%) while pHN was most prominent upon supine head turning (in 80.5%). Theoretically, a symmetrical bilateral p-BPPV could lead to paroxysmal upbeat nystagmus during coactivation of both posterior semicircular canals in the SHH position. For HC-BPPV, the nystagmus that occurs when the left ear or right ear is in the down position can be directed to the undermost (geotropic) or uppermost (apogeotropic) ear (1). Thus, this feature (geotropic or apogeotropic) cannot be used to differentiate a peripheral or central origin.

Only one patient had a positive D–H with the direction of nystagmus characteristic of BPPV (31) but failed to respond to repositioning treatments, and later developed ataxia. Cho and colleagues (26) presented three patients who, except for apogeotropic PN during supine roll tests, had normal neurological examinations. After an initial diagnosis of BPPV, canalith repositioning maneuvers were applied repeatedly but without success, and the patients were later found to have central lesions. These cases highlight the importance of considering central causes of PN when presumed BPPV is refractory to treatment.

Direction-changing nystagmus is described in the literature as nystagmus whose fast phase direction changes (e.g., from right-beating to left-beating) when the position of the head changes (37). Direction-changing pHN on roll test strongly favors the diagnosis of HC-BPPV (7). However, direction-changing nystagmus was noted in 62.5% of CPN participants on D–H (4, 7, 16, 17, 25, 30, 34) and should alert to the presence of CPN during the D–H.

Direction reversing is a different phenomenon, developing spontaneously in two successive phases after the patient reaches a particular position (and without further head movement). It has been suggested that the first phase represents the pathological nystagmus caused by a central vestibular disorder (vestibulocerebellar dysfunction), while the secondary nystagmus represents the adaptive mechanism to nullify the original pathological nystagmus (15). Direction reversing was only reported in three (3.9%) CPN participants and may have a high positive predictive value for central pathology, but relatively low negative predictive value.

Our data set demonstrates a high prevalence of central paroxysmal positional nystagmus (CPPN), which predominantly decays in under 30 s on D–H and SHH. However, CPN triggered by supine head roll typically lasted at least 1 min, mimicking the observations of cupulolithiatic HC-BPPV (1). Choi et al. (4) proposed that the temporal patterns of nystagmus intensity distinguished the two disorders (4).

In our data set, latency proved to be a move reliable indicator of CPN with 94.7% commencing within 3 s (78.9% starting “with no latency”) of assuming the provoking position. Despite the diagnostic strength of this parameter, it was only reported in less than half (46%) of patients within this data set.

The loss of visual suppression of nystagmus with optic fixation indicates a lesion at the flocculonodular lobe of the cerebellum (38, 39). The effect of visual fixation upon nystagmus was only documented in two patients of the papers reviewed here, with patients with posterior fossa lesions demonstrating a failure of fixation suppression (16).