Accuracy of patient-reported spell duration: a comparative study

Abstract

Clinicians rely heavily on patient histories to make medical diagnoses, most of which are inherently subjective and prone to inaccuracies. The aim of this study is to compare the subjective versus objective duration of spells through a retrospective chart review of patients admitted to the epilepsy monitoring unit at our tertiary care medical center. One hundred patients were analyzed. Differences in the accuracy of subjective estimations versus objective duration were compared by age, sex, focal versus generalized, location (frontal versus non-frontal), and spell type (focal aware versus impaired awareness and epileptic versus non-epileptic). Our data show that patients are poor subjective estimators, with 73% of patients overestimating the duration of their spells. We did not find differences in estimated duration by age, sex, seizure location or spell type. A notable exception was patients with generalized convulsive seizures, who accurately reported spell duration to within 17 seconds. This is likely because these seizures are stereotypical, and patients/family time them. Moreover, patients with non-epileptic spells were worse estimators of their spell duration than those with epileptic spells. In addition, although the prefrontal lobe plays a role in time estimation, we did not find that patients with frontal lobe seizures were worse estimators than those with non-frontal seizures, but invasive monitoring can more precisely localize seizures within areas of the frontal lobe responsible for time estimation. Our data emphasize the importance of not relying solely on patient-reported time estimation in diagnosing and developing treatment plans and instead instructing patients to time their spells.

1. Introduction

Neurologists rely on patient histories to make medical diagnoses, yet their histories are inherently subjective. Acquiring an accurate history is particularly important in episodic disorders, for which the duration of the spell is paramount for accurate diagnosis. Examples of such spells are transient ischemic attacks (TIAs), benign paroxysmal positional vertigo (BPPV), and seizures, wherein subsequent management is often guided by the description and duration of symptoms reported by the patient or their caregivers. Three possible factors are thought to affect the estimation accuracy of spell duration. First, spells that are not stereotypical and that vary in duration, such as non-epileptic spells, may make the estimation of their duration less precise [1]. Second, the prefrontal cortex, including the medial frontal lobe, is currently regarded as the cortical region importantly involved in time perception [2], [3]. This implies that the accuracy of time estimation could be affected in disorders involving frontal cortex [2]. Finally, prior studies on the role of aging in time perception have concluded that senescence is associated with reduced accuracy of time estimation, possibly due to frontal lobe degeneration with age [4]. Though these reasons may explain why patients’ estimations might not be accurate, there is a paucity of literature comparing an individual’s perceived estimation of their spell duration (i.e., subjective duration) and the actual duration (i.e., objective duration). One such study analyzed primarily focal aware seizures and found a low correlation between subjective and objective duration of seizures [5]. No other studies have evaluated subjective versus objective time estimation by patients with conditions that could lead them to make inaccurate estimations, such as those with non-epileptic spells, frontal lobe dysfunction, or the elderly. Such a study would advance our understanding of time estimation in different patient populations which, depending on the results, could influence how physicians should elicit patient histories.

The primary aim of our study is to compare the subjective and objective duration of spells (as recorded with video-EEG) by performing a retrospective chart review of patients admitted to our epilepsy monitoring unit (EMU). To better characterize the accuracy of time estimation, we compared differences in the accuracy of subjective estimation versus their measured objective duration. We also aimed to study this accuracy in subgroups by spell type (epileptic versus non-epileptic), location of focal seizures (frontal versus non-frontal), age (65 or older versus younger), type of seizure onset (focal versus generalized), and sex (females versus males).

We hypothesize that patients with spells frequently overestimate the duration. This is based on clinical experience, where patients with spells due to posterior canal BPPV, which objectively last less than one minute, often overestimate the duration of their spells by several minutes or longer [6], [7]. Furthermore, there is evidence that the quality of an event determines the perception of time, where during a pleasant event, time flies (contracts); but if the event is negative/fearful (such as a seizure would be), then time freezes (perceived longer) [8], [9]. In addition, we further hypothesize that some patient characteristics (i.e., older age), spell location (frontal lobe versus non-frontal lobe), and spell type (epileptic versus non-epileptic) also influence accuracy of estimation of duration.

2. Methods

2.1. Patient population and procedure

We performed a retrospective chart review of patients admitted to the EMU at our tertiary medical center, a level IV comprehensive epilepsy program. Our study included adults over the age of 18 years from January 2015 to June 2021. Only EMU admissions that captured at least one spell were included in the study. From the medical charts of the EMU admission, we collected the subjective duration of spells as reported by the patients including those with impaired awareness; only in very few patients was the subjective duration reported in the chart by an eyewitness. We also collected objective duration data from the charts; we did not review the video-EEG recordings. In addition, we collected the following data: age (at the time of EMU admission); sex; handedness; type of onset (focal versus generalized); location of focal seizures (frontal versus non-frontal); and spell type (focal aware versus impaired awareness and epileptic versus non-epileptic). These demographic data constituted our patient subgroups. For patients who had multiple spells during their EMU admission, we only used the first spell that was recorded and that was representative of a typical or habitual spell for that patient. We collected one spell per patient. Our study was approved by our Institutional Review Board, IRB # 202104682.

2.2. Statistical analyses

Summary statistics are provided for the overall study cohort and are stratified by whether a patient’s subjective perception of their spell duration was shorter or longer than the objective duration. Summary statistics are presented as counts and percentages for categorical data, and as medians and interquartile ranges (IQR) for continuous data. Spearman’s correlation was used to detect whether there is a relationship between the subjective and objective durations of the spell, and Spearman’s rho statistic was used to estimate the rank-based measure of association and to find the p value. Wilcoxon rank sum tests were employed to examine whether different strata impact the gap between subjective and objective spell durations, in which the strata were the predictor, and the outcome was the subjective duration minus the objective duration. All p values less than 0.05 were considered statistically significant, and multiple comparison adjustments were not employed due to the exploratory nature of the analysis. All analyses were conducted using R, version 4.1.3.

3. Results

3.1. Demographics

We reviewed the EMU admission records of 158 patients and collected data for 158 spells (one spell per patient). Our baseline demographics were as follows: the median age of our patients was 34 years (IQR 25–47); 88 (56%) were female, and 70 (44%) were male; 128 had handedness data documented (118 [92%] were right-handed). Of the 158 patients, 58 had either subjective or objective spell duration information missing in the chart and were thus excluded from further analysis.

3.2. Patients overestimate the Duration of their Spells

In our sample of 100 spells, we compared subjective versus objective spell duration for each patient (Table 1). We found that the median of the 100 patient-estimated spell durations was 52% greater than the actual objective spell durations, and that 73/100 (73%) of our patients overestimated the duration of their spells.

Table 1.

Subjective versus Objective Spell Duration in Patient Subgroups

Characteristics	Number of Spells	Subjective Spell Duration Median in seconds (IQR)	Objective Spell Duration Median in seconds (IQR)	p-Value1
Overall	100	120 (60–180)	79 (40–119)	<0.001
Spell Type
Epileptic	84	120 (60–180)	81 (42–113)	<0.001
Non-epileptic	16	420 (60–900)	60 (30–135)	0.011
Epileptic Spell Localization
Frontal	12	120 (105–300)	77 (44–106)	0.015
Non-frontal	70	120 (60–160)	81 (43–115)	0.007
Age
65 or older	9	90 (90–240)	36 (32–60)	0.012
Less than 65	91	120 (60–180)	82 (44–120)	<0.001
Epileptic Spell Onset
Focal Onset	77	120 (60–180)	81 (43–116)	0.001
Generalized	7	60 (60–120)	77 (50–93)	0.697
Awareness
Aware	4	90 (90–90)	32 (30–54)	0.278
Not Aware	59	120 (60–180)	83 (49–116)	0.030
Sex
Female	55	120 (60–180)	83 (56–120)	0.020
Male	45	120 (60–160)	60 (36–110)	<0.001

¹Wilcoxon rank sum test

3.3. Patients with Generalized Convulsive Seizures

We had 7 patients with generalized convulsive seizures. Their spell duration estimations were not significantly different than their actual spell durations. Their subjective and objective estimates were within 17 seconds of each other and did not significantly differ, p = 0.697 (Table 1).

3.4. Focal Aware Seizures

We had 4 patients with focal aware seizures. The difference between their subjective and objective durations was not significant, p = 0.278 (Table 1).

3.5. Duration of non-epileptic Spells is Overestimated

We calculated the accuracy of the estimated spell duration using the absolute difference between the subjective and objective duration for each patient (Table 2). Then we compared the accuracy of one subgroup of patients versus another: 1) patients with epileptic spells versus non-epileptic spells; 2) frontal lobe seizures versus non-frontal lobe seizures; 3) older versus younger patients; 4) focal versus generalized onset seizures; and 5) females versus males.

Table 2.

Comparing the Accuracy of Patient Subgroups

Characteristic	Number of Spells = 100 Absolute difference in seconds between subjective and objective spell duration: median (IQR)	p-Value1
Spell Type		0.045
Epileptic	60 (30–142)
Non-Epileptic	100 (58–853)
Location of Focal Seizures		0.906
(data missing on 18 patients where it is not mentioned in the charts if the focal seizure is of frontal onset or non-frontal)
Frontal *	44 (30–98)
Non-Frontal	60 (27–147)
Age		0.962
65 years or older	60 (57–200)
Younger than 65	60 (30–160)
Seizure Onset		0.132
(missing data on 16 patients, where it is not mentioned in the charts if the seizure is focal or generalized)
Generalized	30 (30–40)
Focal	60 (30–150)
Awareness		0.703
Aware	59 (49–60)
Not Aware	60 (25–189)
Sex		0.144
Female	70 (37–180)
Male	50 (20–132)

¹Wilcoxon rank sum test

^*We had two patients whose seizures localized both to the frontal region as well as to another location. We included these patients in the frontal lobe seizure subgroup because our hypothesis is that frontal lobe dysfunction (such as from seizures) affects the ability of a patient to estimate spell duration accurately. Since this low number of 2 patients limited any meaningful statistical analyses, no further analysis on the effect of temporo-frontal lobe seizures on estimation accuracy was performed.

We found that the median subjective duration of patients with non-epileptic spells were 100 seconds different than the objective duration, while the estimations of epileptic patients were different by a median of 60 seconds; i.e., the patients with non-epileptic spells were less accurate at estimating their spell duration than the patients with epileptic spells, p= 0.045 (Table 2).

3.6. Patients with Frontal lobe seizures are not less accurate than those with other seizure locations, similarly for older patients, focal impaired awareness and between sexes

We found no difference in accuracy among the compared subgroups (Table 2). Specifically, we did not find that patients with frontal lobe seizures were significantly less accurate estimators than those with non-frontal seizures, p = 0.906. Similarly, patients 65 years or older were not less accurate than those under the age of 65, p = 0.962; we only had 9 patients who were older than 65, and thus we could not choose a higher cut off for older age. There was no significant difference in the accuracy of estimation between those who had focal aware versus impaired awareness, p = 0.703. Finally, there was no significant difference in the accuracy of estimation between the sexes, p = 0.144. (Table 2).

4. Discussion

Our study showed that most patients overestimate the duration of their spells and are thus poor subjective estimators. In fact, 73% of our patients overestimated the duration of their spells, and the median of patient-estimated spell durations was 52% greater than the actual spell durations. We found that all patient subgroups, except those with generalized convulsive seizures, were inaccurate in estimating their spell duration. This included epileptic and non-epileptic spells, frontal and non-frontal lobe seizures, older and younger patients, focal and generalized onset epileptic spells, and both sexes.

Of all 150 spells, 13% were generalized, which is consistent with the 15–20% reported [10]. We had a total of 14 generalized convulsive seizures for whom we had objective data available, and we found that the average objective duration of generalized convulsive seizures was 93 seconds (67–118), which is consistent with the 1–2 minutes average duration described in the literature [1]. We only had 7 patients with generalized convulsive seizures who had both subjective and objective data; these patients were accurate in their subjective estimations to within 17 seconds of their objective durations. We propose that this is in part due to the stereotypical nature of generalized convulsive seizures [11]. Additionally, patients and their caregivers are typically educated about the importance of timing their seizures and know to call emergency services if a seizure lasts more than 5 minutes [12].

We found that patients with non-epileptic spells were worse estimators than those with epileptic spells, likely due to increased variability among non-epileptic spells [1]. In addition, non-epileptic spells are different from epileptic spells where in the latter it is evident to the patient/eyewitness when the spell occurs and when it stops which helps in estimation of duration.

We postulated that the estimates made by patients with frontal lobe seizures would be less accurate due to the frontal lobe’s role in time estimation [2], but this was not consistent with our data. We used surface EEG, and thus we could not further localize seizures within the various cortical areas of the frontal lobe.

We also hypothesized that the estimates made by older patients would be less accurate than younger patients, due to cognitive decline and associated frontal lobe atrophy with age [4]. We did not find this to be true in our study, possibly due to the small sample size of 9 older patients. Our sample is a young sample, consistent with other studies [5], [13]. Although not surprising that our focal aware seizure patients were accurate, we had a small number of focal aware seizures and thus possibly the small sample size precluded us from detecting a difference. Finally, both sexes were similarly inaccurate.

Our findings are consistent with the literature comparing subjective time estimation in various medical conditions. In epilepsy, Campora et al studied 35 seizure patients; though their focus was different, they did report on subjective and objective durations and found a weak correlation between them [5]. In BPPV of the posterior canal, where the objective duration of a spell is less than one minute, patients often report their symptoms as lasting several minutes or longer [6], [7]. BPPV is underdiagnosed, partly because of prolonged subjective perception of spell duration [7], [14]. Migraine with aura duration also has time as specific clinical criteria for diagnosis, namely 5–60 minutes [15]; therefore, a patient wrongly reporting their aura as lasting longer than that may be misdiagnosed [15]. Similarly, the diagnosis of a TIA is largely based on subjective patient history, and the ability to accurately diagnose a TIA impacts further treatment decisions, because TIA duration is used to stratify future stroke risk [16]. As supported by our data, relying solely on the subjective clinical history may lead to inaccurate diagnosis and management.

5. Limitations

Our study is limited by its retrospective nature, making it prone missing data. 58/158 spells had missing data for either the objective or subjective seizure duration, or both. Seizure medicines are often lowered or stopped during video-EEG study, which could potentially affect duration of spell, particularly for epileptic patients who may get longer seizures. However, if being off their medication in EMU could prolong seizure duration, our results, if anything, are an underestimation of patients’ inaccuracies and remain valid. We did not collect data on comorbidities. It is estimated that less than 10% of patients with epilepsy have intellectual disability/Autism [17]. Therefore, finding that more than 70% of our patients overestimated is unlikely to be fully accounted for by any cognitive limitations that may exist.

6. Future directions

In the future, comparing objective data obtained without lowering or stopping home medications could be done by studying patients who were allowed to continue their home seizure medications in the EMU or who had ambulatory video-EEG study at home. Furthermore, studying patients prospectively and collecting data about subjective and objective durations of multiple habitual spells in EMG may be worth studying. In addition, studying subjects with invasive EEG recording may help better characterize whether seizure location particularly within the prefrontal lobe has a significant impact on estimating spell duration [18]. And studying a larger cohort of older patients versus younger would be needed to assess age effect. We did not collect data on race/ethnicity. There is potential evidence that “in Western cultures, time flows from left to right” [19]. Therefore, studying patients from different cultures, where language is written from right to left or vertically, may be of value. Finally, collecting subjective data solely from witnesses can control for effect of age, and comparing the two subjective experiences (patient and eyewitness) would be of interest, too, because these two activities (watching the movement versus moving) use different cortical processes (visual and motor pathways respectively) [20].

7. Conclusions

With the exception of patients with generalized convulsive seizures who were accurate estimators, all other patients overestimated the duration of their spells, irrespective of age, sex, seizure location, and spell type. Not surprisingly, focal aware seizures may also be accurate estimators, but our numbers were small. Educating patients and their caregivers on the best practice of timing the spell by looking at a watch or recording it with a video will increase the accuracy of estimation and diagnosis. In sum, we reaffirm the adage that time flies when having fun and freezes (perceived longer) during negative valiance events, such as a spell [9].

Highlights Spell Duration.

• Clinicians relay on patients’ histories, which are subjective and can be inaccurate

• In EMU, we compared patients’ subjective spell duration to the objective duration

• In 100 spells, 73% of patients over-estimated the duration of their spells

• There were no differences by age, sex, seizure location or type (epileptic vs none)

• One exception is that patients with generalized seizures had accurate estimation

• This study emphasizes the importance of not relying on patient-reported estimations

• For diagnosis, we recommend to patients to actually time or videotape their spells