Aanvalsdetectie
Uitgangsvraag
In welke situatie is welke aanvalsdetectie gebaseerd op beweging, hartslagfrequentie(variabiliteit), electromyografie, geluid en elektrodermale activiteit voor het detecteren van en alarmeren over epileptische aanvallen bij patiënten met epilepsie geïndiceerd?
Aanbeveling
Overweeg het gebruik van aanvalsdetectieapparatuur bij patiënten met epilepsie van 4 jaar en ouder met tonisch-clonische aanvallen of andere aanvallen met veel motorische verschijnselen en risico’s op aanvalscomplicaties. Aanvalsdetectie heeft vooral een aanvullende waarde als deze aanvallen nu soms gemist worden. Deze leeftijdsgrens wordt gegeven gezien het ontbreken van literatuur over patiënten jonger dan 4 jaar.
Adviseer het gebruik indien er één of meer nachtelijke tonisch-clonische aanvallen per maand optreden als deze (deels) gemist worden, behalve als er praktische bezwaren of andere nadelen zijn.
Maak de keuze wel of geen aanvalsdetectie in overleg met de patiënt en/of zijn of haar familie of verzorgers, waarbij de voor- en nadelen en de doelen van aanvalsdetectie dienen te worden gesproken. Neem hierin ook de haalbaarheid van de opvolging van alarmen mee.
Stem het type aanvalsdetectie zo goed mogelijk af op de specifieke patiënt (situatie). Overweeg bij nachtelijke aanvallen de Nightwatch of Epicare Free of Mobile. Overweeg bij aanvallen overdag de Epicare Free of Mobile.
Bij wens voor aanvalsdetectie overweeg verwijzing naar een van de epilepsiecentra (SEIN of Kempenhaeghe). Zij kunnen inventariseren of er mogelijk bruikbare methoden beschikbaar zijn bij die specifieke patiënt en indien dit het geval is kan de beste methode onderzocht worden.
Overwegingen
Voor- en nadelen van de interventie en de kwaliteit van het bewijs
Aanvalsdetectie apparaten worden beoordeeld op meerdere aspecten, zoals hoe goed werkt het apparaat (sensitiviteit, het aantal fout positieve alarmen en de positieve predictieve waarde), gebruiksgemak, kosten, en ook de beschikbaarheid in Nederland inclusief een CE-markering. Daarnaast zijn de karakteristieken van de gebruikers (patiënt en omgeving) van belang.
Harde grenzen om aan te geven hoe goed een apparaat is zijn er niet, want dit ligt ook aan de wensen en mogelijkheden van de gebruiker. Globaal kan gesteld worden dat een sensitiviteit van 70% en een fout positieve alarm hoeveelheid van 1/dag ondergrenzen zijn. Een sensitiviteit van boven 85% wordt als ‘goed’ beschouwd.
Veel apparatuur voor aanvalsdetectie combineren in de registratie het detecteren van beweging en of hartritme veranderingen, zoals de Nightwatch. De meest gebruikte bewegingssensoren maken gebruik van accelerometrie. De sensitiviteit wisselt tussen de studies van 48% tot 100%. Ook de frequentie fout positieve alarmen varieert, waarbij belangrijk is om te melden dat er in de nacht relatief minder fout positieve alarmen gevonden worden in vergelijking met overdag. Er is ook aanvalsdetectie apparatuur die gebaseerd is op hartslag en beweging, zoals de Epicare Mobile/Free. Er zijn aanwijzingen dat bij correct gebruik de sensitiviteit goed is. De frequentie van fout positieve alarmen varieert ook bij dit type aanvalsdetectie.
De gevonden sensitiviteit bij detectie gebaseerd op autonome respons zweten alleen was matig.
Drie studies onderzochten detectiemethoden gebaseerd op verschillende modaliteiten (beweging + electrodermale activiteit en beweging + hartfrequentie). De gevonden sensitiviteit was hoog.
Op één studie na (Arends, 2018) zijn er geen studies gevonden die aanvalsdetectiemethoden met elkaar vergelijken. Op basis van de wetenschappelijke literatuur is het, in zijn algemeenheid, niet te zeggen welke aanvalsdetectie-apparaat het meest accuraat epileptische aanvallen detecteert. Dit zal in de praktijk ook afhangen van de semiologie van de aanvallen.
Volgens de GRADE-systematiek werd laag tot zeer laag bewijs gevonden voor alle detectie-apparatuur, onder andere vanwege beperkingen in de studie-opzet en de lage patiënt aantallen. De GRADE systematiek lijkt echter niet het meest geschikt voor het beoordelen van studies naar monitoring apparatuur. Overigens werd een deel van de studies (gedeeltelijk) gefinancierd door de fabrikant van het betreffende apparaat. Bij een aantal van deze studies was een deel van de auteurs medewerker of anderzijds belanghebbende van de fabrikant (zie tabellen).
Na de zoekdatum is in 2022 een studie gepubliceerd die aanvalsdetectie gebaseerd op beweging en hartfrequentie, de Nightwatch, onderzocht bij 23 kinderen in de thuissituatie (Lazeron, 2022). De sensitiviteit was redelijk tot hoog (overall sensitiviteit 79,4%; mediane sensitiviteit per patiënt 93,2%). De frequentie fout positieve alarmen was hoger dan in een eerdere studie bij volwassenen waarbij de fout positieve alarmen voornamelijk optraden als de kinderen wakker werden. Na aanpassing van het algoritme waarbij alarmen enkel nog optraden als de kinderen in horizontale positie lagen, was de frequentie fout positieve alarmen lager met behoud van een vergelijkbare sensitiviteit.
Er is indirect bewijs dat nachtelijk toezicht de kans op SUDEP kan verminderen (Sveinsson, 2020; van der Lende, 2018). Data uit een case-controle studie uit Zweden liet zien dat het SUDEP risico voor patiënten met tonisch clonische aanvallen aanmerkelijk hoger is als zij alleen slapen. Het hebben van tonisch clonische aanvallen verhoogt het SUDEP risico met aan factor 18 voor patiënten die samen slapen, maar is 67 (of bijna 70) keer hoger als iemand alleen slaapt (Sveinson, 2020). Een onderzoek op een begeleid wonen afdeling gradeerde de mate van nachtelijk toezicht in drie categorieën: het risico op SUDEP (na correctie voor aanvalsfrequentie) was het hoogst in de groep zonder enig toezicht; terwijl deze het laagst was in de groep met het meeste toezicht. Deze laatste categorie bestond uit mensen bij wie er naast uitluisteren ook andere maatregelen getroffen werd om toezicht te vergroten (niet meer alleen slapen, frequente nachtelijke controles, video bewaking of inzet aanvalsdetectieapparatuur) (van der Lende, 2018)
De gevonden variatie tussen alle studies is onder andere afhankelijk van het type aanvallen; tonisch-clonische aanvallen worden beter gedetecteerd dan aanvallen zonder of met minder motorische verschijnselen. Het type epilepsie is daarbij vermoedelijk niet van invloed maar niet onderzocht. Bovendien zijn de sensitiviteit en fout positieve alarmen afhankelijk van de ingestelde grenswaarden, zoals de minimale duur van de aanvallen die gedetecteerd kunnen worden. De algoritmen die gebruikt worden zijn complex; dit kan de variatie in de resultaten deels verklaren. Vrijwel alle studies zijn verricht op epilepsie monitoring units, er zijn vooralsnog slechts weinig studies gedaan in de thuis- of woonsituatie, de zogenoemde fase 4 studies (Beniczky, 2017), zoals bijvoorbeeld onderzoek van Arends (2018).
Voor de dagelijkse praktijk worden vooral de Nightwatch (aanvallen ’s nachts) en EpiCare Free / Mobile (aanvallen overdag en ’s nachts) geadviseerd vanwege de aspecten zoals hierboven genoemd (accuraatheid, beschikbaarheid voor de Nederlandse markt). In praktijk worden de Emfit en Embrace ook als mogelijkheden gezien, ondanks minder harde wetenschappelijk bewijzen maar vanwege soms goede gebruikerservaringen. Andere apparatuur, bij andersoortige aanvallen is buiten beschouwing gelaten, evenals op EEG gebaseerde monitoring.
Voor kinderen onder de 4 jaar is er geen specifieke aanvalsdetectie onderzocht. Bij aanvallen met een daling in de zuurstof saturatie of een duidelijke verandering in de hartslag kan een saturatiemeter overwogen worden.
Waarden en voorkeuren van patiënten (en evt. hun verzorgers)
De meeste aanvalsdetectie-apparaten zijn in de vorm van een bandje om de pols of arm. Over het algemeen worden deze goed verdragen, maar net name bij kinderen en patiënten met een verstandelijke beperking kan het dragen van zo’n bandje bezwaarlijk zijn. Belangrijk is verder rekening te houden met eventuele fout-positieve alarmen en dit ook te bespreken. De impact van fout-positieve alarmen op patiënten, hun familie of begeleiders is persoonlijk en onder andere afhankelijk van de setting (thuissituatie of zorginstelling), de aanvalsfrequentie en de belastbaarheid van het gezin (van Westrheenen, 2021) . Ook kunnen bijkomende bewegingsstoornissen de aanvalsdetectie door beweging geassocieerde detectie-apparatuur hinderen.
Indien gewenst kan voor de inventarisatie van de meest geschikte aanvalsdetectie verwezen worden naar de gespecialiseerde poliklinieken van de epilepsiecentra Kempenhaeghe en SEIN. Aldaar kan bij individuele patiënten, geselecteerd op type aanvallen, mate van voorkomen en andere karakteristieken, middels onderzoek met geselecteerde aanvalsdetectoren nagegaan worden of deze goed bruikbaar zijn. Daarbij wordt dan ook een beeld gekregen van de sensitiviteit, de frequentie van fout-positieve alarmen en gemiste aanvallen bij de individuele patiënt zodat een specifieker afweging van nut en noodzaak gemaakt kan worden.
Informatie over aanvalsdetectiemethoden valt ook te lezen op de website https://epilepsie.nl/aanvalsdetectie.
Kosten (middelenbeslag)
Er is nog weinig bekend over de kosteneffectiviteit van aanvalsdetectie. Er is een onderzoek verschenen na de zoekdatum (Engelgeer, 2022) die de suggestie geeft dat het implementeren van de Nightwatch kosteneffectief is. De onderzoekspopulatie betrof thuiswonende kinderen met focale epilepsie en tenminste een aanval per week. Momenteel wordt aanvalsdetectieapparatuur niet vergoed vanuit de basisverzekering.
Aanvaardbaarheid, haalbaarheid en implementatie
Met name in de nacht, als er een minder goede observatie mogelijk is, kan het detecteren van epileptische aanvallen zinvol zijn. Dit geldt niet voor elke patiënt, maar vooral voor degene met frequente of ernstige aanvallen waarbij er een hoog risico is op aanvalscomplicaties zoals verwondingen of SUDEP. Een ander argument voor aanvalsdetectie bestaat als het wenselijk is het aantal aanvallen objectief te registreren.
Onderbouwing
Achtergrond
Aanvalsdetectie wordt voornamelijk gebruikt om te alarmeren bij het optreden van een epileptische aanval zodat iemand uit de omgeving naar de patiënt kan gaan en kan ingrijpen indien nodig. Dit kan overwogen worden bij iedereen met onbehandelbare epileptische aanvallen die hoge gezondheidsrisico’s zoals sudden unexpected death in epilepsy (SUDEP), status epilepticus en verwondingen met zich meebrengen, voornamelijk tonisch-clonische aanvallen of andere aanvallen met veel motoriek. Het heeft vooral aanvullende waarde als de aanvallen zonder aanvalsdetectie gemist worden, zoals aanvallen in de nacht als patiënten alleen slapen.
Daarnaast zouden betrouwbare aanvalsdetectie apparaten gebruikt kunnen worden om het aantal aanvallen objectief te registreren. Het is belangrijk om te weten hoe vaak iemand een epileptische aanval heeft, onder andere voor de behandeling. Met name ook door te registreren op momenten die anders mogelijk onopgemerkt zouden blijven (bijvoorbeeld in de nacht) zullen apparaten die betrouwbaar aanvallen kunnen registreren nauwkeuriger zijn dan het bijhouden van een aanvalsdagboek.
Er zijn veel verschillende (ambulante) methodes van aanvalsdetectie, dit is de laatste jaren sterk in ontwikkeling. Beschikbare systemen zijn bijvoorbeeld accelerometers die bewegingen en lichaamshouding registreren, matrassensoren, electromyografie om de spieractiviteit te meten, ECG om de hartfrequentie en variabiliteit hierin te meten, geluidsdetectiesystemen en systemen die electrodermale activiteit, zuurstofsaturatie of respiratie meten. In deze module wordt onderzocht welke aanvalsdetectiesysteem het meest geschikt zijn.
Conclusies
1. Seizure detection based on motor response
| Low GRADE | The evidence suggests that wrist-worn seizure detection based on motor response may have a good sensitivity but poor positive predictive value in detecting seizures in patients with epilepsy in a health care setting.
Sources: Benickzy, 2013; Johansson, 2019; Kramer, 2011; Nijsen, 2015; Velez, 2016 |
| Very low GRADE | The evidence is very uncertain about the effect of using EMG surface electrodes on biceps for detecting epileptic seizures in a health care setting.
Source: Szabo, 2015 |
2. Seizure detection based on autonomic response
| Very low GRADE | The evidence is very uncertain about the effect of using a neurostimulation devise equipped with a cardiac based seizure detection algorithm for detecting epileptic seizures in a health care setting.
Source: Boon, 2015 |
| Very low GRADE | The evidence is very uncertain about the effect of electrodes recording ECG signals (on lower left ribs) for detecting epileptic seizures in a health care setting.
Source: Jeppesen, 2020 |
3. Seizure detection based on multimodal response
| Low GRADE | The evidence suggests that wrist-worn seizure detection based on electrodermal activity and accelerometry may have a good sensitivity but poor positive predictive value in detecting seizures in patients with epilepsy in a health care setting.
Source: Onorati, 2017; Onorati, 2021 |
| Low GRADE | The evidence suggests that seizure detection based on heartrate and accelerometry may have a good sensitivity but poor positive predictive value in detecting seizures in patients with refractory epilepsy in a residential care setting.
Source: Arends, 2018 |
Samenvatting literatuur
Six studies describe seizure detection systems based on motor response (table 1), three studies based on autonomic response (table 2), 1 study based on noise response and three study based on multimodal responses (table 4). We have not found eligible studies based on indirectly motor response.
Table 1: Motor response
| Study reference | Study characteristics | Seizure detection | Patient characteristics | Device of interest | Reference standard (R) | Follow-up | Outcome measures and effect size | Comments |
| Nijsen, 2005 | Type of study: Observational study
Setting: Long-stay environment
Country: The Netherlands
Funding: Not reported
Conflict of interest: Not reported
| Seizure detection feature: Accelerometry
Type of seizure: Motor seizures | Inclusion criteria: Patients mentally retarded suffer from severe epilepsy Minimum seizure frequency 20/month.
Exclusion criteria: 2 pts excluded (1 no seizures, 1 technical difficulties with EEG)
Number of patients: 20 patients were recruited from whom 18 patients had 897 seizures
Age: Mean age ± SD: 37 ± 11,8 years
Sex: 56% M / 44% F
| Describe index test: 3D accelerometers created by 2 2D sensors (ADXL202E from Analog Devices Inc) at perpendicular angles to each other. Data stored on portable recorders. Afterward recordings moved to analyzing system (Brainlab). Monitoring 36 hours
Cut-off point(s): Not reported
| Describe reference test: v-EEG
Cut-off point(s): EEG technicians detect seizures using 2 possible paradigms: 1) first screen video recordings for behavioral information that corresponds to a seizure and then additionally check EEG signal for epileptiform activity. 2) first check EEG signal and then screen video recordings for behavioral information | Time between the index test and reference test: Same time.
For how many participants were no complete outcome data available? 2 patients
Reasons for incomplete outcome data described? 1 patient did not have any seizure (within 36 hours study period) 1 due to technical difficulties | Outcome measures and effect size (include 95%CI and p-value if available):
Sensitivity: 47.7%
PPV: Not applicable, false positive seizures not reported | Classification based on Benickzy & Ryvlin: Phase 1 or 2 (>10pts, offline recordings, analysis not real time, no alarms. Safety of device not adressed)
|
| Kramer, 2011 | Type of study: Prospective observational study
Setting: 2 v-EEG units
Country: Israel, USA
Funding: None
Conflict of interest: All authors are connected to the company (Biolert Ltd) that develops the device described in this article (founder, leading scientist, employee and consultant). | Seizure detection feature: Accelerometry
Type of seizure: Primarily motor seizures incl. clonic, tonic, partial, generalized
Complex partial seizures | Inclusion criteria: Patients with primarily motor seizures incl. clonic, tonic, partial, generalized.
Patients with complex partial seizures who had generalized seizures after tapering AED.
Exclusion criteria: Patients with dystonic posturing, subtle behavioural automatism, suspected pseudoseizures.
N of patients: 31 patients were recruited from whom 15 patients had 22 seizures
Age: Mean age ± SD: Not reported
Sex: Not reported | Describe index test: ‘Movement Sensor’ device (bracelet) attached to limb. Contains a 3axis accelerometer and transmitter. Transmitted tot computer. System monitored and detected seizures based on algorithm based on various movements’ parameters.
Cut-off point(s): N/A | Describe reference test: Video-EEG. At 1 unit each alarm was immediately checked and confirmed by the nurses
Cut-off point(s): Not applicable | Time between the index test and reference test: Same time
For how many participants were no complete outcome data available? 16 patients (52%) had no complete data available
Reasons for incomplete outcome data described? - Absence of attacks during monitoring - Pts with during monitoring only type of seizure not detected by device (ie complex partial seizures) - Communication failure in the system - Wearing device on non-involved limb | Outcome measures and effect size (include 95%CI and p-value if available):
Sensitivity: 90.9% (95% CI: 72.2-97.5%)
PPV: 71.4% (95% CI: 52.9-84.8%)
Satisfaction: Not reported
| Classification based on Benickzy & Ryvlin: Phase 2 or 3 (>20pts, 15 seizures, online, alarms, multicenter, blinding?) |
| Benickzy, 2013 | Type of study: Prospective observational study
Setting: EMU
Country: Denmark, Germany
Funding: None
Conflict of interest: None | Seizure detection feature: Accelerometry
Type of seizure: GTCS
| Inclusion criteria: Patients at risk of having GTCS
Exclusion criteria: Not described
N of patients: 73 patients were recruited from whom 20 patients had 39 seizures (GTCS)
Age: Mean: 37 years Range: 6-68 years
Sex: 53% M / 47% F | Describe index test: Epi-Care Free, developed by Danish Care Technology ApS. Resembles wristwatch. Contains the three-axis acceleration transducer, microprocessor, and battery. The sensor measures the acceleration of any movement in the wrist. The device has two-way wireless communication to a table placed or portable control unit. A predefined, generic convulsive seizure detection algorithm was used for all monitoring/detections. Realtime calculations determined whether movements were seizure-like or normal. Alarm was triggered at a fixed threshold. The algorithm was fully automatic
Cut-off point(s): Not applicable | Describe reference test: v-EEG
Cut-off point(s): Not applicable | Time between the index test and reference test: Same time.
For how many participants were no complete outcome data available? None of the patients were excluded from analysis, however device deficiency was recorded 15 times
Reasons for incomplete outcome data described? Device deficiency due to technical errors and battery failure, however these problems were corrected and recording was continued | Outcome measures and effect size (include 95%CI and p-value if available):
Sensitivity: 89.7% (95% CI: 76.4-95.9%)
PPV: 46.7% (95% CI: 35.8-57.8%)
Satisfaction: mild allergic reaction on the skin recorded in 1 patient | Classification based on Benickzy & Ryvlin: Phase 3 study (continuous recording, prospective, real time. Predefined algorithms & cutoff values, multicenter, blinded, >30 seizures recorded) |
| Szabó, 2015 | Type of study: Prospective observational study
Setting: EMU
Country: USA
Source of funding: funded by Brain Sentinel Inc.
Conflict of interest: One author is the President, CEO, and shareholder of Brain Sentinel Inc., and another author (last author) is a consultant and shareholder of Brain Sentinel, Inc
| Seizure detection feature: EMG
Type of seizure: Medically refractory epilepsy and history of GTCS
| Inclusion criteria: Patients with medically refractory epilepsy and history of GTCS
Exclusion criteria: N/A
Number of patients: 36 patients were recruited from whom 11 patients had 21 seizures (GTCS)
Age: Mean: 40 Range: 14-64 years
Sex: 48% M / 52% F | Describe index test: Surface EMG electrodes on m. biceps & m. triceps on the arm suspected primarily or mostly involved. Raw EMG signal recorded using conventional amplifiers. Only biceps sEMG recordings utilized for analysis, Elektrodes connected to NeXus-32 or NeXus-4 monitor (Mind Media BV), transmitted to portable laptop or exported using Biotrace+.
Cut-off point(s): at onset 2 measurements of maximum voluntary biceps contraction to set the baseline physiological muscle contraction threshold. sEMG signal analyzed using frequency based algorithm developed by Brain Sentinel, fully automated. | Describe reference test: v-EEG
Cut-off point(s): N/A | Time between the index test and reference test: Same time
For how many participants were no complete outcome data available? 3 patients (8%) had no complete data available
Reasons for incomplete outcome data described? Yes, 2 patients had faulty recordings, 1 patient EMG recordings not retrievable. | Outcome measures and effect size (include 95%CI and p-value if available):
Sensitivity: 95.2% (95% CI: 77.3-99.1%)
PPV: 95.2% (95% CI: 77.3-99.2%) | Classification based on Benickzy & Ryvlin:
Classification of Study: Phase 2 (>10pts, >15 seizures, offline, no alarms, blinding?) |
| Velez, 2016 | Type of study: Prospective observational study
Setting: EMU
Country: USA
Funding: Wrist accelerometers and tablets were loaned to study group by SmartMonitor©
Conflicts of interest One co-author has stock options in SmartMonitor
| Seizure detection feature: ACM
Type of seizure GTCS
| Inclusion criteria: Patients ≥ 18 years, with in history of tonic-clonic movement in ≥ 1 limb
Exclusion criteria: Only non-tonic-clonic or suspected psychogenic non-epileptic events
N of patients: 30 patients were recruited from whom 27 patients had 62 seizures (13 GTCS and 49 non-GTCS)
Mean age ± SD: Not reported. Range 19-66 years
Sex: 66,7% M / 33,3% F
| Device of interest SmartWatch: Wireless wristwatch biosensor programmed to detect shaking events and record the date, start &end times, duration, mean shake frequency and amplitude of abnormal movements and associated audio
Cut-off point(s): Thresholds for shake detections were set to a sensitivity of 4 on a scale of 10
Comparator test: Bedside paper diary by patients / caregivers
Cut-off point(s): N/A | Describe reference test: v-EEG
Cut-off point(s): Seizure intensity: Mild was defined as a myoclonic, tremor, tonic or clonic movement in one limb, moderate as movements in more than one limb but without full body convulsions and severe as full body or generalized tonic-clonic movements
Frequency: low (10 Hz).
Predominant amplitude: visually categorized into low, medium or high | Time between index test and reference test: Same time
For how many participants were no complete data available? 3 participants had no complete data available.
Reasons for incomplete outcome data described? Yes, 1 for early discharge, 1 v-EEG was not recorded at time of seizure, 1 lost watch before seizure | Outcome measures and effect size (include 95%CI and p-value if available):
Sensitivity: 92.3% (95%CI: 66.7-98.6%)
False negative: 1 seizure
PPV: 24.0% (95%CI: 14.3-37.4%)
Satisfaction: Not reported
| Classification based on Benickzy & Ryvlin: Phase 2 (or 3) (>20 pts, >30 seizures. Dedicated device. Continues recording. Online database, but analysis after the recording. Not multicenter)
Other comments: Event data were successfully uploaded to the bedside tablet in 11/12 (91.7%) and to the online database in 10/12 (83.3%) of the GTCS.
The watch recorded 81 false positives, of these 42 (51.8%) were cancelled by patients
|
| Johansson, 2019 | Type of study: Prospective observational study
Setting: Hospital
Country: Sweden
Funding: Swedish Foundation for Strategic Research from Sahlgrenska Academy University of Gothenburg
Conflicts of interest: 2 co-authors employees of RISE Acreo AB, 2 co-authors MSs students at RISE ACREo AB
| Seizure detection feature: Accelerometry
Type of seizure: Tonic-clonic seizures
| Inclusion criteria: Epilepsy surgery candidates > 18 years
Exclusion criteria: Not described
N of patients 75 patients were recruited from whom 11 patients had 37 TCSs
Age: Median: 35 years Range: 18-77 years
Sex 27% M / 63% F
Other important characteristics History of TCSs: 81% | Devices of interest Two types of inertial wrist-worn sensors during different phases of data collection. Initial phase: RISE Acreo (in-house developed) Later phase: Shimmer3 (Shimmer Research Ireland)
Cut-off point(s): 0.1g movement threshold to remove non-motor epochs
| Describe reference test: Seizures identified by epileptologist, based on video EEG
Cut-off point(s): Not reported
| Time between index test and reference test: Same time
For how many participants were no complete data available? 29 participants had no complete data available
Reasons for incomplete outcome data described? Yes, no sensors on, sensors were removed due to discomfort, technical errors and unknown reasons
| Outcome measures and effect size (include 95%CI and p-value if available):
Sensitivity: 82.2% 95%CI: 68.7-90.7%)
PPV: 66.1% (95%CI: 53.0-77.1%)
Satisfaction: Not reported
| Classification based on Benickzy & Ryvlin Phase 2 |
Table 2: Autonomic response
| Study reference | Study characteristics | Seizure detection | Patient characteristics | Device of interest | Reference standard (R) | Follow -up | Outcome measures and effect size | Comments |
| Boon, 2015 | Type of study: Prospective multicenter observational study
Setting: EMU
Country: Belgium, Germany, The Netherlands, Norway, United Kingdom
Funding: Cyberonics, Inc.
Conflicts of interest: 4 co-authors employees at Cyberonics Inc. Contributions included design of the trial, data acquisition, data analysis, interpretation of data, and drafting manuscript.
| Seizure detection feature: Cardiac
Type of seizure: Tonic-clonic
| Inclusion criteria: Patients ≥ 18 years, VNS candidates with history of iCT
Exclusion criteria: Not reported N of patients 31 patients were recruited from whom 16 patients had 66 seizures
5 patients with 11 seizures associated with iTC
Mean age: 39.6 years ± SD 13.4 years
Sex: 38.7% M / 61.3 % F
| Device of interest: AspireSR device. Patients were implanted with the neurostimulation devise equipped with novel CBSDA. Provides an automatic stimulation feature which is triggered in response to ictal heart rate increases of at least 20% to deliver VNS in a closed-loop fashion.
After EMU discharge (ambulatory setting), patients underwent both open- and closed-loop VNS.
Cut-off point(s): iCT defined as ictal heart rate > 100 bpm, and at least 55% increase, or 35 bpm increase from baseline. Patients randomized to three different SDA thresholds (≥20%, ≥ 40%, ≥60% above baseline heart rate)
| Describe reference test: Continuous v-EEG and ECG monitoring was performed during 3 – 5 days when the device was programmed with the closed-loop VNS feature only during the EMU stay. Ambulatory phase
Cut-off point(s): Not reported | Time between index test and reference test: Same time
For how many participants were no complete data available? 1 participant had no complete data available
Reasons for incomplete outcome data described? 1 patient discontinued the trial prior to the 12 month follow-up visit due to an adverse event of diarrhea and vomiting | Outcome measures and effect size (include 95%CI and p-value if available):
Based on observed analysis of the realworld behavior of the device:
iTC seizures: 11 in 5 patients Overall sensitivity: 81.8% (9/11) PPV: N/A Sensitivity per threshold: - ≥20%: 100% (2/2) - ≥ 40%: 75% (6/8) - ≥60%: 100% (1/1)
Seizures with ≥ 20% heart rate change Overall sensitivity: 56.8% (21/37) Sensitivity per threshold: - ≥20%: 100% (11/11) - ≥ 40%: 46.7% (7/15) - ≥60%: 27.3% (3/11)
Satisfaction: Patient: N/A Caregiver: N/A Professional: Adverse effects: Dysphonia in 9 patients (29%) and diarrhea and vomiting in 1 patient (3.2%)
| Classification based on Benickzy & Ryvlin Phase 2
Other comments: - After EMU discharge, patients underwent both open- and closed-loop VNS. - QOL results over a 12-month treatment period: Responder rate of 30%. Following domains in QOL exceeded the MIC criteria for clinical significance: emotional wellbeing, social function, cognitive function, seizure worry and overall QOL by combined closed- and open-loop VNS. No numbers reported.
|
| Jeppesen, 2020 | Type of study: Prospective multicenter observational study
Setting: Epilepsy centre and university hospital
Country: Denmark
Funding: ePatch wearable ECG devices provided by FORCE Technology company
Conflicts of interest: None
| Seizure detection feature: HRV
Type of seizure: Convulsive and nonconvulsive seizures
| Inclusion criteria: Patients undergoing surgical or diagnostic evaluation
Exclusion criteria: Patients who had pacemaker, VNS, or any implanted electronic device
N of patients 47 patients were recruited from whom 19 patients had 48 seizures.
Responders: 11 patients (57.9%)
Non-responders: 8 patients (42.1%)
Age Median: 27 years Range 4-62 years
Sex 51.1% M / 48.9% F
| Device of interest: ePatch. Self-adhesive patch containing electrodes recording ECG signals. Placed on the lower left ribs. Custom-made computer programs developed in LabVIEW 2016 (64-bit) (National Instruments) calculated HRV parameters from all ECG data.
Cut-off point(s): Responders: First seizure >50 bpm HR change. Based on the following criteria: patients who had two-thirds or more of their seizures detected based on HRV. Positive detection when HRV parameters exceeded the individualized cut-off values (set to 105% of the highest baseline values) in period from 120 heartbeats before seizure onset to seizure termination. Individualized cut-off values determined by: Heart rate differential method (HR-diff), Cardiac Sympathetic Index (CSI), modified CSI (ModCSI), and if applicable bike exercise test (completed by 11 patients (57.9%)) and/or cognitive test (completed by 14 patients (73.7.%)).
Best seizure detection algorithm studied earlier by Jeppesen (2019) was used.
Non-responders: patients who did not meet above mentioned criteria.
| Describe reference test: v-EEG
Cut-off point(s): Not reported | Time between index test and reference test: Same time. Algorithm was run offline, not real-time
For how many participants were no complete data available? 2 participants had no complete data avaiiable
Reasons for incomplete outcome data described? Yes, two patients were excluded due to poor skin-to-electrode connection of the adhesive tape of the ePatch | Outcome measures and effect size (include 95%CI and p-value if available):
Among responders and nonresponders combined: N=48 seizures Sensitivity: 56.3% (95%CI: 42.2-70.3) False negative: 21 seizures PPV: N/A
Among responders N= 23 seizures Sensitivity: 87.0% (95% CI: 73.2-100%) PPV: 86.2% (95% CI: 73.7-98.8)
Among non-responders N=25 seizures Sensitivity: 28.0% (95% CI: 12.1-49.4%) PPV: N/A
Satisfaction: Not reported
| Classification based on Benickzy & Ryvlin Phase 2 |
Table 3: Multimodal response
| Study reference | Study characteristics | Seizure detection | Patient characteristics | Device of interest | Reference standard (R) | Follow -up | Outcome measures and effect size | Comments |
| Onorati, 2017 | Type of study: Retrospective multicenter observational study
Setting: EMU
Country: USA & Italy
Funding: Not reported
Conflict of interest: Seven out of 19 authors are employees of the company Empatica that developed the device. Another author is part of pending patent applications to detect and predict seizures and to diagnose epilepsy with devices different from the ones used in this work. | Seizure detection feature: Electrodermal and accelerometry
Type of seizures: Convulsive seizures (FTC and FTCb)
| Inclusion criteria: Patients with epilepsy admitted for v-EEG
Exclusion criteria: Not reported.
Number of patients: 69 patients were recruited from whom 22 patients had 55 seizures.
Age: Children: 24 patients Median: 14 years Range: 4-18 years
Adults: 45 patients Median: 37 years Range: 19-60 years
Sex: Children: 62,5% M / 37.5% F
Adults: 37,8% M / 62,2% F
| Devices of interest: 3 different wristbands recording electrodermal activity&accelerometer signals (E3 or E4 (Empatica) or iCalm (MIT Media Lab), at wrist where convulsions appeared earlier/more evident, otherwise nondominant arm. 5pts on both wrists. 3 different feature sets extracted on each epoch.
Cut-off point(s): 3 different feature sets extracted. Dataset splitted into 3 parts: 2 parts for training, 1 for testing. 3x, whole dataset evaluated. Optimal decision threshold selected by using FAR/Sens ROC curves. | Describe reference test: v-EEG
Cut-off point(s): Not reported | Time between the index test and reference test: Same time
For how many participants were no complete outcome data available? 0 patients | Outcome measures and effect size (include 95%CI and p-value if available):
Differs for 3 classifiers. Most efficient classifier: Sensitivity: 94.6% (95% CI: 85.2-98.1%)
PPV: 50.1% (95% CI: 41.4-60.5%)
Satisfaction: Not reported
| Classification based on Benickzy & Ryvlin: Phase 2 (>20pts, >15 seizures, offline, retrospective analysis, training&testing same dataset |
Arends, 2018
| Type of study: Prospective observational study
Setting: In-home, residential care setting
Country: The Netherlands
Funding: Grant from the Dutch National Science Foundation (NWO-ZonMW) No. 300040003, the NUTS-Ohra Foundation No. 1203-050, and the Dutch Epilepsy Foundation for extra material costs. LivAssured, the company developing the Nightwatch device, has obtained an exclusive license to implement or use the data in the future for commercial purposes or in commercial enterprises in exchange for a percentage of the revenue for the institutes.
Conflicts of interest: Authors are members of the Dutch Tele-Epilepsy Consortium. None of the authors has shares in the funding company / institutes, nor has any of them (will) received compensation referring to future sales of the Nightwatch. The Dutch Tele-Epilepsy Consortium will receive more research funds from the institutes as a consequence of this license.
| Seizure detection feature: Plethysmography and ACC
Type of seizures: Major nocturnal seizure
| Inclusion criteria: Adults with intellectual disability, history of >1 major nocturnal seizure per month & resided in a long-term facility of a participating epilepsy center
Exclusion criteria: Patients with movement disorder, pacemaker or, skin pigmentation
Number of patients: 28 patients had 809 major seizures.
Age: Mean: 29.1 years Range: 15-67 years
Sex: 64.3% M / 35.7% F
| Describe index test: Nightwatch, developed by LivAssured BV. Bracelet worn around upper arm, measuring HR (by plethysmography) and movement (3D ACC). Signals and online alarms wirelessly transmitted to base station connected to computer that also was connected to an infrared-sensitive video camera
Cut-off point(s): Alarm generated only if position indicates lying and threshold for HR (slope of tachycardia) exceeded or motion value stayed above threshold ≥ 15 sec.
Comparison: Measured in 14 patients | Describe reference test: In addition, in a random sample of 10% of all nights, the complete video recordings were screened to estimate the number of missed major seizures. Cut-off point(s): Video event was considered true positive when alarm was given within 3 minutes before or 5 minutes after start of a major seizure. | Time between the index test and reference test: Same time
For how many participants were no complete outcome data available? 6 patients had no complete data available
Reasons for incomplete outcome data described? Yes, 2 patients were not motivated, 1 did not accept video, 1 did not accept Nightwatch, 2 had no major seizures
| Outcome measures and effect size (include 95%CI and p-value if available):
Median sensitivity per participant: 86% (95%CI: 77 - 93%).
Median PPV per participant: 49% (95%CI: 33-64%)
MedianPPV per participant 49%
Satisfaction: 33 caregivers completed questionnaire.
With regard to care - Helps caregiver provide better care: 22 yes, 5 no, 5 other, 1 n/a or no answer - Helps relieve burden of caregiving: 7 caregivers yes, 26 no
With regard to sensor: - Accepted by patient: 28 yes, 2 no, 2 other, 1 n/a - Problems with sensor: 13 yes, 8 no, 12 n/a or no answer
With regard to usability: Sufficient technical support: 27 yes, 1 no, 5 other
With regard to autonomy: - Offers more autonomy to people with epilepsy: 10 yes, 13 no, 10 n/a or no answer - Offers caregiver more freedom: 20 yes, 11 no, 2 n/a or no answer
User-friendliness 7.3 on a scale from 1-10 (1 lowest user-friendliness) Privacy 4.2 (1 being lowest level of invasion)
Outcome measures and effect size of comparison (include 95%CI and p-value if available): Overall sensitivity 21% (95%CI: 6 - 32%).
| Classification based on Benickzy & Ryvlin: Phase 2 (≥10 patients, ≥ 15 seizures)
Other comment: Nightwatch compared with piezoelectric Emfit bed sensor by 14 patients. Median difference 58% (95%CI: 39-80%)
Index test is part of reference standard |
Onorati, 2021
| Type of study Prospective multicenter clinical study
Setting: EMU
Country: USA, Italy
Funding: grants from Epilepsy Foundation, Norman Prince Neurosciences Institute, Brown Institute for Brain Sciences, Epilepsy Research Foundation, American Epilepsy Society, Patient-Centered Outcomes Research Institute (PCORI), Pediatric Epilepsy Research Foundation, Citizens United for Research in Epilepsy (CURE) Foundation, HHV-6 Foundation, Lundbeck, Eisai Ltd, Upsher-Smith Inc., Acorda Therapeutics Inc., and Pfizer Inc.
Conflicts of interest: First (and corresponding author) is employee at Empatica, Inc., Boston | Seizure detection feature: ACC and EDA
Type of seizures: CS (FBTC and GTC)
| Inclusion criteria: Children ≥6 -≤ 20 years and adults ≥21 years.
Exclusion criteria: Not reported.
N of patients: 304 patients from whom 152 patients’ datasets were analyzed.
Seizures: Total: 36 patients had 66 convulsive seizures Children: n=18 with 35 convulsive seizures Adults: n= 18 with 31 convulsive seizures
Age: Children: 85 patients Median: 12 years Range: 6-20 years
Adults: 67 patients Median: 38 years Range: 21-63 years
Sex: Children: 55.3% M / 44.7% F
Adults: 44.8% M / 55.2% F
| Devices of interest Two multimodal wrist-worn devices: 1) E4 embeds ACC-, EDA-, PPG-, and temperature sensor 2) Embrace embeds ACC, gyroscope, EDA, and temperature sensor
Cut-off point(s): Only recorded ACM and EDA data acquired by the two devices were used for detection algorithm. Tow steps: dataset. i.e., a collection of labeled sensor data, was processed to obtain a set of features on windowed sensor data. Same procedure for separate validation datasets. After defining performance metrics to maximize, labeled features from training dataset were provided to machine learning algorithm to obtain classification model and decision rule function. Both devices considered as equivalent for ACM and EDA sensor data and therefore used interchangeably in the study (E4 worn by 124 (81.6%) patients and Embrace by 28 patients (18.4%))
| Describe reference test: v-EEG
Cut-off point(s): Not reported | Time between index test and reference test: Same time
For how many participants were no complete data available? 152 participants had no complete data available
Reasons for incomplete outcome data described? Yes, 112 patients did not wear wrist sensor device, 1 patient did not have prior epilepsy diagnosis, 22 patients were < 6 years old, 10 patients were excluded due to hardware, software or data issues of the EEG reference device or the wearable device (4 patients), 1 patient due to lack of compliance, and 2 patients dropping out from the study | Outcome measures and effect size (include 95%CI and p-value if available):
‘FDA-cleared’ mode (children and adults): Sensitivity: 98.5% (95% CI 91.9-99.7%) or cSensitivity: 96% (95% CI: 92-100) False negative: 1 PPV: 15.4 (95% CI: 12.3-19.2%)
Satisfaction: Not reported
| Classification based on Benickzy & Ryvlin Phase 3 |
Abbreviations:
ACC Accelerometry; ACM Accelerometer; CBSDA cardiac based seizure detection algorithm; CS Convulsive seizures; ECG Electrocardiography; EDA Electrodermal activity; EMU Epilepsy monitoring unit; FBTC Focal onset to bilateral/unilateral tonic-clonic; FDA US Food and Drug Administration; FTC Focal motor tonic-clonic; FTCb Focal motor to bilateral tonic-clonic; GTC Generalized onset tonic-clonic; GTCS Generalized tonic-clonic seizures; HRV Heart rate variability; iCT Ictal tachycardia; MIC Minimally important change; PPG Photoplethysmography; PPV Positive predicted value; QOL quality of life; SDA Seizure detection algorithm; v-EEG Video-EEG; VNS Vagus nerve stimulation
Conclusion
The grading system used is probably not completely suitable for monitoring devices. For this specific situation, standards for diagnostic methods (that require to make a reliable assessment on a single assessment) or therapeutic interventions (that require a proof of improvement in outcome compared to no treatment) do not fully apply, yet elements such as risk of bias and imprecision can be judged in a similar manner. In this type of research relatively low patient numbers are not easily to overcome. For practical purpose the principles of GRADE diagnostic are used. The level of evidence for starts high. The level of evidence was downgraded to moderate, low or very low certainty in case of risk of bias, inconsistency, indirectness, imprecision, or publication bias.
1. Seizure detection based on motor response
Wrist-worn seizure detection: the level of evidence was downgraded by one level due to study limitations (-1, risk of bias, see risk of bias tables) and downgraded by one level due to imprecision (low number of study participants).
3D-accelerometer/EMG surface electrodes on biceps: the level of evidence regarding was downgraded by one level due to study limitations (-1, risk of bias/publication bias) and downgraded by two levels due to imprecision (low number of study participants).
2. Seizure detection based on autonomic response
Neurostimulation device with a cardiac based seizure detection algorithm / electrodes recording ECG signals (on lower left ribs): the level of evidence was downgraded by one level due to study limitations (-1, risk of bias/publication bias) and downgraded by two levels due to imprecision (low number of study participants).
3. Seizure detection based on multimodal response
Wristbands recording electrodermal activity and accelerometry: the level of evidence was downgraded by one level due to study limitations (-1, risk of bias/publication bias) and downgraded by one level due to imprecision (low number of study participants).
Wrist-worn seizure detection based on heart rate and accelerometry: the level of evidence regarding was downgraded by one level due to study limitations (-1, risk of bias, see risk of bias tables) and downgraded by one level due to imprecision (low number of study participants).
Zoeken en selecteren
A systematic review of the literature was performed to answer the following question:
What is the accuracy of seizure detection systems based on motor response, indirect motor response, autonomic response, and seizure detection systems based on noise from patients to detect and alert epileptic seizures among patients with different types of epilepsy.
PICRO
P: Adults and children with epilepsy
I: Motor response:
Accelerometry, gyroscope, magnetometers
ElectroMyoGraphic
Indirect motor response:
Video, infrared video system, radar
Bed mattress, floor mat or other changes by movement
Noise by movement
Autonomic response:
Heart rate (variability) (electrocardiography, photoplethysmography, video, radar)
To sweat (electrodermal activity)
Other, e.g. breathing, oxygen saturation
Noises from patient (noise detection system or direct listening)
(C: another seizure detection system)
R: Video and electroencephalography or video alone
O: Accuracy: Sensitivity, false-negative rate, and positive predicted value.
Satisfaction (patient, caregiver, professional)
Relevant outcome measures
The guideline development group considered accuracy measures as critical outcome measures for decision making; and satisfaction (patient, caregiver, professional) as an important outcome. The accuracy consists of sensitivity, number of false positive and negative alarms.
Satisfaction was only described in the literature summary if the device had at least ‘reasonable’ accuracy for detection epileptic seizures.
The working group found it hard to define thresholds for sensitivity. Globally, a sensitivity between 70% to 84% was considered ‘reasonable’ and above 85% was considered ‘good’ sensitivity.
Search and select (Methods)
The databases Pubmed/Medline, and Embase (via Ovid) were searched with relevant search terms from 01-01-2020 (last search) until 10-05-2022. The detailed search strategy is depicted under the tab Methods. The systematic literature search resulted in 109 hits. Studies were selected based on the following criteria:
(1) RCTs, cohort studies, case-control studies or systematic reviews;
(2) full-text English language publication;
(3) according to PICO;
(4) level 2 or higher according to classification Benickzy&Ryvlin (2018)
Four studies were included since the last update (see the table with reasons for exclusion under the tab Methods),
Results
Thirteen number of studies were included in the analysis of the literature. The assessment of the risk of bias is summarized in the risk of bias tables. The studies were grouped in seizure detection based on motor response, autonomic response and multimodal response (see Table 1-3).
Referenties
- Arends, J., Thijs, R. D., Gutter, T., Ungureanu, C., Cluitmans, P., Van Dijk, J., ... & Dutch Tele-Epilepsy Consortium. (2018). Multimodal nocturnal seizure detection in a residential care setting: a long-term prospective trial. Neurology, 91(21), e2010-e2019.
- Beniczky, S., Polster, T., Kjaer, T. W., & Hjalgrim, H. (2013). Detection of generalized tonic–clonic seizures by a wireless wrist accelerometer: a prospective, multicenter study. Epilepsia, 54(4), e58-e61.
- Beniczky, S., & Ryvlin, P. (2018). Standards for testing and clinical validation of seizure detection devices. Epilepsia, 59, 9-13.
- Boon, P., Vonck, K., van Rijckevorsel, K., El Tahry, R., Elger, C. E., Mullatti, N., ... & McGuire, R. M. (2015). A prospective, multicenter study of cardiac-based seizure detection to activate vagus nerve stimulation. Seizure, 32, 52-61.
- Engelgeer, A., van Westrhenen, A., Thijs, R. D., & Evers, S. M. (2022). An economic evaluation of the NightWatch for children with refractory epilepsy: Insight into the cost-effectiveness and cost-utility. Seizure, 101, 156-161.
- Jeppesen, J., Fuglsang-Frederiksen, A., Johansen, P., Christensen, J., Wüstenhagen, S., Tankisi, H., ... & Beniczky, S. (2020). Seizure detection using heart rate variability: a prospective validation study. Epilepsia, 61, S41-S46.
- Johansson, D., Ohlsson, F., Krýsl, D., Rydenhag, B., Czarnecki, M., Gustafsson, N., ... & Malmgren, K. (2019). Tonic-clonic seizure detection using accelerometry-based wearable sensors: a prospective, video-EEG controlled study. Seizure, 65, 48-54.
- Kramer, U., Kipervasser, S., Shlitner, A., & Kuzniecky, R. (2011). A novel portable seizure detection alarm system: preliminary results. Journal of Clinical Neurophysiology, 28(1), 36-38.
- Lazeron, R. H., Thijs, R. D., Arends, J., Gutter, T., Cluitmans, P., Van Dijk, J., ... & Dutch Tele‐Epilepsy Consortium. (2022). Multimodal nocturnal seizure detection: Do we need to adapt algorithms for children?. Epilepsia Open, 7(3), 406-413.
- Nijsen, T. M., Arends, J. B., Griep, P. A., & Cluitmans, P. J. (2005). The potential value of three-dimensional accelerometry for detection of motor seizures in severe epilepsy. Epilepsy & Behavior, 7(1), 74-84.
- Onorati, F., Regalia, G., Caborni, C., Migliorini, M., Bender, D., Poh, M. Z., ... & Picard, R. W. (2017). Multicenter clinical assessment of improved wearable multimodal convulsive seizure detectors. Epilepsia, 58(11), 1870-1879.
- Onorati, F., Regalia, G., Caborni, C., LaFrance Jr, W. C., Blum, A. S., Bidwell, J., ... & Picard, R. (2021). Prospective study of a multimodal convulsive seizure detection wearable system on pediatric and adult patients in the epilepsy monitoring unit. Frontiers in Neurology, 1444.
- Sveinsson, O., Andersson, T., Mattsson, P., Carlsson, S., & Tomson, T. (2020). Clinical risk factors in SUDEP: a nationwide population-based case-control study. Neurology, 94(4), e419-e429.
- Szabó, C. Á., Morgan, L. C., Karkar, K. M., Leary, L. D., Lie, O. V., Girouard, M., & Cavazos, J. E. (2015). Electromyography‐based seizure detector: Preliminary results comparing a generalized tonic–clonic seizure detection algorithm to video‐EEG recordings. Epilepsia, 56(9), 1432-1437.
- van der Lende, M., Hesdorffer, D. C., Sander, J. W., & Thijs, R. D. (2018). Nocturnal supervision and SUDEP risk at different epilepsy care settings. Neurology, 91(16), e1508-e1518.
- van Westrhenen, A., de Lange, W. F., Hagebeuk, E. E., Lazeron, R. H., Thijs, R. D., & Kars, M. C. (2021). Parental experiences and perspectives on the value of seizure detection while caring for a child with epilepsy: A qualitative study. Epilepsy & Behavior, 124, 108323
- Velez, M., Fisher, R. S., Bartlett, V., & Le, S. (2016). Tracking generalized tonic-clonic seizures with a wrist accelerometer linked to an online database. Seizure, 39, 13-18.
Evidence tabellen
Risk of bias assessment diagnostic accuracy studies (QUADAS II, 2011)
Research question:
| Study reference | Patient selection
| Index test | Reference standard | Flow and timing | Comments with respect to applicability |
|---|---|---|---|---|---|
| Nijsen, 2005 | Was a consecutive or random sample of patients enrolled? Yes
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Yes
| Were the index test results interpreted without knowledge of the results of the reference standard? Unclear
If a threshold was used, was it pre-specified? Yes
| Is the reference standard likely to correctly classify the target condition? Yes
Were the reference standard results interpreted without knowledge of the results of the index test? Unclear
| Was there an appropriate interval between index test(s) and reference standard? Yes
Did all patients receive a reference standard? Yes
Did patients receive the same reference standard? Yes
Were all patients included in the analysis? No 2 patients were excluded: 1 patient had no seizures during study period, 1 patient due to technical difficulties | Are there concerns that the included patients do not match the review question? No
Are there concerns that the index test, its conduct, or interpretation differ from the review question? No
Are there concerns that the target condition as defined by the reference standard does not match the review question? No
|
CONCLUSION: Could the selection of patients have introduced bias?
RISK: LOW | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: UNCLEAR
| CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias? RISK: UNCLEAR | CONCLUSION Could the patient flow have introduced bias?
RISK: HIGH | ||
| Kramer, 2011 | Was a consecutive or random sample of patients enrolled? Yes
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Yes
| Were the index test results interpreted without knowledge of the results of the reference standard? No
If a threshold was used, was it pre-specified? Unclear
| Is the reference standard likely to correctly classify the target condition? Yes
Were the reference standard results interpreted without knowledge of the results of the index test? No
| Was there an appropriate interval between index test(s) and reference standard? Yes
Did all patients receive a reference standard? Yes
Did patients receive the same reference standard? Yes
Were all patients included in the analysis? No, 16 patients (52%) excluded: absence of attacks during monitoring, patients with during monitoring only type of seizure not detected by device (ie complex partial seizures), communication failure in the system and, wearing device on non-involved limb
| Are there concerns that the included patients do not match the review question? No
Are there concerns that the index test, its conduct, or interpretation differ from the review question? No
Are there concerns that the target condition as defined by the reference standard does not match the review question? No
|
CONCLUSION: Could the selection of patients have introduced bias?
RISK: LOW | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: UNCLEAR
| CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias? RISK: UNCLEAR | CONCLUSION Could the patient flow have introduced bias?
RISK: HIGH | ||
| Benickzy, 2013 | Was a consecutive or random sample of patients enrolled? Yes
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Yes
| Were the index test results interpreted without knowledge of the results of the reference standard? Yes
If a threshold was used, was it pre-specified?
| Is the reference standard likely to correctly classify the target condition? Yes
Were the reference standard results interpreted without knowledge of the results of the index test? Yes
| Was there an appropriate interval between index test(s) and reference standard? Yes
Did all patients receive a reference standard? Yes
Did patients receive the same reference standard? Yes
Were all patients included in the analysis? Yes | Are there concerns that the included patients do not match the review question? No
Are there concerns that the index test, its conduct, or interpretation differ from the review question? No
Are there concerns that the target condition as defined by the reference standard does not match the review question? No
|
CONCLUSION: Could the selection of patients have introduced bias?
RISK: LOW | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: LOW
| CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias?
RISK: LOW | CONCLUSION Could the patient flow have introduced bias?
RISK: LOW | ||
| Szabó, 2015 | Was a consecutive or random sample of patients enrolled? Yes
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Yes
| Were the index test results interpreted without knowledge of the results of the reference standard? Unclear. sEMG signal was fully automated analyzed offline using a frequency-based algorithm, with the exception of setting the power threshold for the initial matrix calculation, which required the operator to select raw EMG data collected during an MVC (maximum voluntary biceps contraction) and a quiet period.
If a threshold was used, was it pre-specified? Yes
| Is the reference standard likely to correctly classify the target condition? Yes
Were the reference standard results interpreted without knowledge of the results of the index test? Unclear
| Was there an appropriate interval between index test(s) and reference standard? Yes
Did all patients receive a reference standard? Yes
Did patients receive the same reference standard? Yes
Were all patients included in the analysis? Yes | Are there concerns that the included patients do not match the review question? No
Are there concerns that the index test, its conduct, or interpretation differ from the review question? No
Are there concerns that the target condition as defined by the reference standard does not match the review question? No
Brain Sentinel Inc. funded the study. One author is the President, CEO, and shareholder, and another author (last author) is a consultant and shareholder of Brain Sentinel, Inc |
CONCLUSION: Could the selection of patients have introduced bias?
RISK: LOW | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: UNCLEAR
| CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias?
RISK: LOW | CONCLUSION Could the patient flow have introduced bias?
RISK: LOW | ||
| Velez, 2016 | Was a consecutive or random sample of patients enrolled? Yes
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Yes
| Were the index test results interpreted without knowledge of the results of the reference standard? No (fellowship-trained epileptologist compared watch detected online entries to vEEG)
If a threshold was used, was it pre-specified? Yes
| Is the reference standard likely to correctly classify the target condition? Yes
Were the reference standard results interpreted without knowledge of the results of the index test? No (fellowship-trained epileptologist compared watch detected online entries to vEEG)
| Was there an appropriate interval between index test(s) and reference standard? Yes
Did all patients receive a reference standard? Yes
Did patients receive the same reference standard? Yes
Were all patients included in the analysis? No, 3 patients excluded: 1 for early discharge, 1 v-EEG was not recorded at time of seizure, and 1 lost watch before seizure | Are there concerns that the included patients do not match the review question? No
Are there concerns that the index test, its conduct, or interpretation differ from the review question? No
Are there concerns that the target condition as defined by the reference standard does not match the review question? No
The equipment for the study including the wrist accelerometers and tablets were loaned to Stanford University by SmartWatch, SmartMonitor |
CONCLUSION: Could the selection of patients have introduced bias?
RISK: LOW | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: HIGH
| CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias?
RISK: HIGH | CONCLUSION Could the patient flow have introduced bias?
RISK: LOW | ||
| Johansson, 2019 | Was a consecutive or random sample of patients enrolled? Yes
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Yes
| Were the index test results interpreted without knowledge of the results of the reference standard? Yes
If a threshold was used, was it pre-specified? Yes
| Is the reference standard likely to correctly classify the target condition? Yes
Were the reference standard results interpreted without knowledge of the results of the index test? Yes
| Was there an appropriate interval between index test(s) and reference standard? Yes
Did all patients receive a reference standard? Yes
Did patients receive the same reference standard? Yes
Were all patients included in the analysis? No, in 29 patients missing data: no sensors on, removing sensors due to discomfort, unrecorded by sensors due to technical errors and unknown reasons. | Are there concerns that the included patients do not match the review question? No
Are there concerns that the index test, its conduct, or interpretation differ from the review question? No
Are there concerns that the target condition as defined by the reference standard does not match the review question? No
|
CONCLUSION: Could the selection of patients have introduced bias?
RISK: LOW | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: LOW | CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias?
RISK: LOW | CONCLUSION Could the patient flow have introduced bias?
RISK: HIGH | ||
| Boon, 2015 | Was a consecutive or random sample of patients enrolled? Yes
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Yes
| Were the index test results interpreted without knowledge of the results of the reference standard? No, for the observed analysis, timestamps from detection logs that were downloaded from the implanted generator were compared to the seizure annotations from the physicians based on EEG
If a threshold was used, was it pre-specified? Yes
| Is the reference standard likely to correctly classify the target condition? Yes
Were the reference standard results interpreted without knowledge of the results of the index test? Yes
| Was there an appropriate interval between index test(s) and reference standard? Yes
Did all patients receive a reference standard? Yes
Did patients receive the same reference standard? Yes
Were all patients included in the analysis? No, 1 patient discontinued the trial prior to the 12-month follow-up visit due to adverse event.
| Are there concerns that the included patients do not match the review question? Yes, only 5 patients with 11 iCT seizures match to the primairy aim of the research question. According to classification based on Benickzy & Ryvlin: phase 2 number of patients with seizures ≥10 and number of seizures ≥ 15
Are there concerns that the index test, its conduct, or interpretation differ from the review question? No
Are there concerns that the target condition as defined by the reference standard does not match the review question? No
|
CONCLUSION: Could the selection of patients have introduced bias?
RISK: LOW | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: LOW
| CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias?
RISK: LOW | CONCLUSION Could the patient flow have introduced bias?
RISK: LOW | ||
| Jeppesen, 2019 | Was a consecutive or random sample of patients enrolled? Yes
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Yes
| Were the index test results interpreted without knowledge of the results of the reference standard? Unclear, algorithms were run retrospective (offline) on the dataset collected prospectively
If a threshold was used, was it pre-specified? Yes
| Is the reference standard likely to correctly classify the target condition? Yes
Were the reference standard results interpreted without knowledge of the results of the index test? Yes
| Was there an appropriate interval between index test(s) and reference standard? Yes
Did all patients receive a reference standard? Yes
Did patients receive the same reference standard? Yes
Were all patients included in the analysis? No, 1 patient excluded due to poor skin connection of adhesive tape
| Are there concerns that the included patients do not match the review question? No
Are there concerns that the index test, its conduct, or interpretation differ from the review question? No
Are there concerns that the target condition as defined by the reference standard does not match the review question? No
|
CONCLUSION: Could the selection of patients have introduced bias?
RISK: LOW | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: LOW? (thresholds used) | CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias?
RISK: LOW | CONCLUSION Could the patient flow have introduced bias?
RISK: LOW | ||
| Jeppesen, 2020 | Was a consecutive or random sample of patients enrolled? Yes
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Yes
| Were the index test results interpreted without knowledge of the results of the reference standard? Unclear, analysis was done offline, using the predefined algorithm and individualized (patient-tailored) cutoff values
If a threshold was used, was it pre-specified? Yes
| Is the reference standard likely to correctly classify the target condition? Yes
Were the reference standard results interpreted without knowledge of the results of the index test? Yes
| Was there an appropriate interval between index test(s) and reference standard? Yes
Did all patients receive a reference standard? Yes
Did patients receive the same reference standard? Yes
Were all patients included in the analysis? No, two patients were excluded due to poor skin-to-electrode connection of the adhesive tape of the ePatch
| Are there concerns that the included patients do not match the review question? No
Are there concerns that the index test, its conduct, or interpretation differ from the review question? No
Are there concerns that the target condition as defined by the reference standard does not match the review question? No
|
CONCLUSION: Could the selection of patients have introduced bias?
RISK: LOW | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: LOW
| CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias?
RISK: LOW | CONCLUSION Could the patient flow have introduced bias?
RISK: LOW | ||
| Onorati, 2017 | Was a consecutive or random sample of patients enrolled? Unclear
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Unclear
| Were the index test results interpreted without knowledge of the results of the reference standard? No
If a threshold was used, was it pre-specified? No, 3 different feature sets were used. On each set a supervised machine learning classifier was built Optimal decision threshold selected by using FAR/Sens ROC curves
| Is the reference standard likely to correctly classify the target condition? Yes
Were the reference standard results interpreted without knowledge of the results of the index test? Yes
| Was there an appropriate interval between index test(s) and reference standard? Yes
Did all patients receive a reference standard? Yes
Did patients receive the same reference standard? Yes
Were all patients included in the analysis? Yes
| Are there concerns that the included patients do not match the review question? No
Are there concerns that the index test, its conduct, or interpretation differ from the review question? No
Are there concerns that the target condition as defined by the reference standard does not match the review question? No
|
CONCLUSION: Could the selection of patients have introduced bias?
RISK: UNCLEAR | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: HIGH?
| CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias?
RISK: LOW | CONCLUSION Could the patient flow have introduced bias?
RISK: LOW | ||
| Arends, 2018 | Was a consecutive or random sample of patients enrolled? Unclear
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Unclear why 6 patients from the first phase of the study not continued to the second phase of the study.
| Were the index test results interpreted without knowledge of the results of the reference standard? Probably yes
If a threshold was used, was it pre-specified? Yes
| Is the reference standard likely to correctly classify the target condition? Unclear
Were the reference standard results interpreted without knowledge of the results of the index test? No | Was there an appropriate interval between index test(s) and reference standard? n.a.
Did all patients receive a reference standard? Video images of all reported events (from devices, nurses’ records, diaries, and notes) were annotated by the trial nurses. In addition, in a random sample of 10% of all nights, the complete video recordings were screened to estimate the number of missed major seizures.
Did patients receive the same reference standard? See above.
Were all patients included in the analysis? See inappropriate exclusions
| Are there concerns that the included patients do not match the review question? Some doubts; persons with a movement disorder, pacemaker, or skin pigmentation were excluded.
Are there concerns that the index test, its conduct, or interpretation differ from the review question? No
Are there concerns that the target condition as defined by the reference standard does not match the review question? No
|
CONCLUSION: Could the selection of patients have introduced bias?
RISK: UNCLEAR | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: LOW
| CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias?
RISK: UNCLEAR | CONCLUSION Could the patient flow have introduced bias?
RISK: HIGH | “The multimodal sensor was compared with the bed sensor in 14 participants who already used the bed sensor before the start of the trial.”
“Periods of poor signal quality were excluded by the algorithm.” Since not all video material was reviewed, it seems that seizures during this time were not included?
| |
Onorati, 2021
| Was a consecutive or random sample of patients enrolled? Yes
Was a case-control design avoided? Yes
Did the study avoid inappropriate exclusions? Yes
| Were the index test results interpreted without knowledge of the results of the reference standard? Unclear
If a threshold was used, was it pre-specified? Yes
| Is the reference standard likely to correctly classify the target condition? Yes
Were the reference standard results interpreted without knowledge of the results of the index test? Yes
| Was there an appropriate interval between index test(s) and reference standard? Yes
Did all patients receive a reference standard? Yes
Did patients receive the same reference standard? Yes
Were all patients included in the analysis? Yes/No/Unclear.
| Are there concerns that the included patients do not match the review question? Yes/No/Unclear
Are there concerns that the index test, its conduct, or interpretation differ from the review question? Yes/No/Unclear
Are there concerns that the target condition as defined by the reference standard does not match the review question? Yes/No/Unclear
|
CONCLUSION: Could the selection of patients have introduced bias?
RISK: LOW | CONCLUSION: Could the conduct or interpretation of the index test have introduced bias?
RISK: LOW
| CONCLUSION: Could the reference standard, its conduct, or its interpretation have introduced bias?
RISK: LOW | CONCLUSION Could the patient flow have introduced bias?
RISK: LOW /HIGH/UNCLEAR |
Judgments on risk of bias are dependent on the research question: some items are more likely to introduce bias than others, and may be given more weight in the final conclusion on the overall risk of bias per domain:
Patient selection:
- Consecutive or random sample has a low risk to introduce bias.
- A case control design is very likely to overestimate accuracy and thus introduce bias.
- Inappropriate exclusion is likely to introduce bias.
Index test:
- This item is similar to “blinding” in intervention studies. The potential for bias is related to the subjectivity of index test interpretation and the order of testing.
- Selecting the test threshold to optimise sensitivity and/or specificity may lead to overoptimistic estimates of test performance and introduce bias.
Reference standard:
- When the reference standard is not 100% sensitive and 100% specific, disagreements between the index test and reference standard may be incorrect, which increases the risk of bias.
- This item is similar to “blinding” in intervention studies. The potential for bias is related to the subjectivity of index test interpretation and the order of testing.
Flow and timing:
- If there is a delay or if treatment is started between index test and reference standard, misclassification may occur due to recovery or deterioration of the condition, which increases the risk of bias.
- If the results of the index test influence the decision on whether to perform the reference standard or which reference standard is used, estimated diagnostic accuracy may be biased.
- All patients who were recruited into the study should be included in the analysis, if not, the risk of bias is increased.
Judgement on applicability:
Patient selection: there may be concerns regarding applicability if patients included in the study differ from those targeted by the review question, in terms of severity of the target condition, demographic features, presence of differential diagnosis or co-morbidity, setting of the study and previous testing protocols.
Index test: if index tests methods differ from those specified in the review question there may be concerns regarding applicability.
Reference standard: the reference standard may be free of bias but the target condition that it defines may differ from the target condition specified in the review question.
Table of excluded studies
| Reference | Reason for exclusion |
| van Westrhenen A, De Cooman T, Lazeron RHC, Van Huffel S, Thijs RD. Ictal autonomic changes as a tool for seizure detection: a systematic review. Clin Auton Res. 2019 Apr;29(2):161-181. doi: 10.1007/s10286-018-0568-1. Epub 2018 Oct 30. PMID: 30377843; PMCID: PMC6459795. | Studies in SR considered as phase 0 or 1 studies according to classification Benickzy&Ryvlin). Only one study included in summary of literature: Boon, 2015 |
| Beniczky S, Wiebe S, Jeppesen J, Tatum WO, Brazdil M, Wang Y, Herman ST, Ryvlin P. Automated seizure detection using wearable devices: A clinical practice guideline of the International League Against Epilepsy and the International Federation of Clinical Neurophysiology. Epilepsia. 2021 Mar;62(3):632-646. doi: 10.1111/epi.16818. PMID: 33666944. | Wrong study design (no original literature) |
| Regalia G, Onorati F, Lai M, Caborni C, Picard RW. Multimodal wrist-worn devices for seizure detection and advancing research: Focus on the Empatica wristbands. Epilepsy Res. 2019 Jul;153:79-82. doi: 10.1016/j.eplepsyres.2019.02.007. Epub 2019 Feb 27. PMID: 30846346. | Wrong study design (no original literature) |
| van Westrhenen A, Petkov G, Kalitzin SN, Lazeron RHC, Thijs RD. Automated video-based detection of nocturnal motor seizures in children. Epilepsia. 2020 Nov;61 Suppl 1(Suppl 1):S36-S40. doi: 10.1111/epi.16504. Epub 2020 May 7. Erratum in: Epilepsia. 2021 Nov;62(11):2881. PMID: 32378204; PMCID: PMC7754425. | Does not meet PICRO (algorithm) |
| Hegarty-Craver M, Kroner BL, Bumbut A, DeFilipp SJ, Gaillard WD, Gilchrist KH. Cardiac-based detection of seizures in children with epilepsy. Epilepsy Behav. 2021 Sep;122:108129. doi: 10.1016/j.yebeh.2021.108129. Epub 2021 Jun 17. PMID: 34147021; PMCID: PMC8429110. | Does not meet PICRO (algorithm performance) |
| Ryvlin P, Beniczky S. Seizure detection and mobile health devices in epilepsy: Recent developments and future perspectives. Epilepsia. 2020 Nov;61 Suppl 1:S1-S2. doi: 10.1111/epi.16702. Epub 2020 Oct 23. PMID: 33098105. | Wrong study design (brief report) |
| Rheims S. Wearable devices for seizure detection: Is it time to translate into our clinical practice? Rev Neurol (Paris). 2020 Jun;176(6):480-484. doi: 10.1016/j.neurol.2019.12.012. Epub 2020 Apr 28. PMID: 32359805. | Wrong study design (overview article) |
| Beniczky S, Wiebe S, Jeppesen J, Tatum WO, Brazdil M, Wang Y, Herman ST, Ryvlin P. Automated seizure detection using wearable devices: A clinical practice guideline of the International League Against Epilepsy and the International Federation of Clinical Neurophysiology. Epilepsia. 2021 Mar;62(3):632-646. doi: 10.1111/epi.16818. PMID: 33666944. | double |
| Ong JS, Wong SN, Arulsamy A, Watterson JL, Shaikh MF. Medical Technology: A Systematic Review on Medical Devices Utilized for Epilepsy Prediction and Management. Curr Neuropharmacol. 2022;20(5):950-964. doi: 10.2174/1570159X19666211108153001. PMID: 34749622. | Most studies in SR do not meet PICRO (15 articles EEG systems, 2 studies small study population, 2 studies algorithm,1 study development system). One study Velez 2016 included. Two other eligible studies (Boon 2015 en Johansson 2019) already included in summary. |
| Shum J, Friedman D. Commercially available seizure detection devices: A systematic review. J Neurol Sci. 2021 Sep 15;428:117611. doi: 10.1016/j.jns.2021.117611. Epub 2021 Aug 6. PMID: 34419933. | Does not meet PICRO (wrong outcomes) |
| Shum J, Friedman D. Commercially available seizure detection devices: A systematic review. J Neurol Sci. 2021 Sep 15;428:117611. doi: 10.1016/j.jns.2021.117611. Epub 2021 Aug 6. PMID: 34419933. | Does not meet PICRO (algorithm) |
| Yang Y, Sarkis RA, Atrache RE, Loddenkemper T, Meisel C. Video-Based Detection of Generalized Tonic-Clonic Seizures Using Deep Learning. IEEE J Biomed Health Inform. 2021 Aug;25(8):2997-3008. doi: 10.1109/JBHI.2021.3049649. Epub 2021 Aug 5. PMID: 33406048. | Does not meet PICRO (algorithm) |
| Maguire MJ, Jackson CF, Marson AG, Nevitt SJ. Treatments for the prevention of Sudden Unexpected Death in Epilepsy (SUDEP). Cochrane Database Syst Rev. 2020 Apr 2;4(4):CD011792. doi: 10.1002/14651858.CD011792.pub3. PMID: 32239759; PMCID: PMC7115126. | Does not meet PICRO (studies to sensitivity of devices were excluded in SR) |
| Bruno E, Biondi A, Böttcher S, Lees S, Schulze-Bonhage A, Richardson MP; RADAR-CNS Consortium. Day and night comfort and stability on the body of four wearable devices for seizure detection: A direct user-experience. Epilepsy Behav. 2020 Nov;112:107478. doi: 10.1016/j.yebeh.2020.107478. Epub 2020 Sep 28. PMID: 33181896. | Accuracy as crucial outcome not measured |
| [51] Frankel MA, Lehmkuhle MJ, Watson M, Fetrow K, Frey L, Drees C, Spitz MC. Electrographic seizure monitoring with a novel, wireless, single-channel EEG sensor. Clin Neurophysiol Pract. 2021 May 29;6:172-178. doi: 10.1016/j.cnp.2021.04.003. PMID: 34189361; PMCID: PMC8220094. [55] Swinnen L, Chatzichristos C, Jansen K, Lagae L, Depondt C, Seynaeve L, Vancaester E, Van Dycke A, Macea J, Vandecasteele K, Broux V, De Vos M, Van Paesschen W. Accurate detection of typical absence seizures in adults and children using a two-channel electroencephalographic wearable behind the ears. Epilepsia. 2021 Nov;62(11):2741-2752. doi: 10.1111/epi.17061. Epub 2021 Sep 7. PMID: 34490891; PMCID: PMC9292701. | Does not meet PICRO (EEG excluded from intervention) |
| Nasseri M, Pal Attia T, Joseph B, Gregg NM, Nurse ES, Viana PF, Worrell G, Dümpelmann M, Richardson MP, Freestone DR, Brinkmann BH. Ambulatory seizure forecasting with a wrist-worn device using long-short term memory deep learning. Sci Rep. 2021 Nov 9;11(1):21935. doi: 10.1038/s41598-021-01449-2. PMID: 34754043; PMCID: PMC8578354. | Wrong study (phase 0 or 1 according to classification Benickzy&Ryvlin). |
| Stirling RE, Grayden DB, D'Souza W, Cook MJ, Nurse E, Freestone DR, Payne DE, Brinkmann BH, Pal Attia T, Viana PF, Richardson MP, Karoly PJ. Forecasting Seizure Likelihood With Wearable Technology. Front Neurol. 2021 Jul 15;12:704060. doi: 10.3389/fneur.2021.704060. PMID: 34335457; PMCID: PMC8320020. | Does not meet PICRO (no use of reference standard) |
| Elezi L, Koren JP, Pirker S, Baumgartner C. Automatic seizure detection and seizure pattern morphology. Clin Neurophysiol. 2022 Jun;138:214-220. doi: 10.1016/j.clinph.2022.02.027. Epub 2022 Mar 18. PMID: 35382982. | Does not meet PICRO (algorithm) |
| Faria MT, Rodrigues S, Campelo M, Dias D, Rego R, Rocha H, Sá F, Tavares-Silva M, Pinto R, Pestana G, Oliveira A, Pereira J, Cunha JPS, Rocha-Gonçalves F, Gonçalves H, Martins E. Heart rate variability in patients with refractory epilepsy: The influence of generalized convulsive seizures. Epilepsy Res. 2021 Dec;178:106796. doi: 10.1016/j.eplepsyres.2021.106796. Epub 2021 Oct 26. PMID: 34763267. | Does not meet PICRO (influence of generalized convulsive seizures in heart rate variability (HRV) in patients with refractory epilepsy) |
| Faria MT, Rodrigues S, Campelo M, Dias D, Rego R, Rocha H, Sá F, Tavares-Silva M, Pinto R, Pestana G, Oliveira A, Pereira J, Cunha JPS, Rocha-Gonçalves F, Gonçalves H, Martins E. Does the type of seizure influence heart rate variability changes? Epilepsy Behav. 2022 Jan;126:108453. doi: 10.1016/j.yebeh.2021.108453. Epub 2021 Dec 1. PMID: 34864377. | Does not meet PICRO (impact of the type of seizure on HRV) |
| Halimeh M, Yang Y, Sheehan T, Vieluf S, Jackson M, Loddenkemper T, Meisel C. Wearable device assessments of antiseizure medication effects on diurnal patterns of electrodermal activity, heart rate, and heart rate variability. Epilepsy Behav. 2022 Apr;129:108635. doi: 10.1016/j.yebeh.2022.108635. Epub 2022 Mar 9. PMID: 35278938. | Does not meet PICRO (effect of antiseizure medications on different data modalities simultaneously recorded by a wearable device) |
| Hossain MB, Posada-Quintero HF, Kong Y, McNaboe R, Chon KH. A Preliminary Study on Automatic Motion Artifact Detection in Electrodermal Activity Data Using Machine Learning. Annu Int Conf IEEE Eng Med Biol Soc. 2021 Nov;2021:6920-6923. doi: 10.1109/EMBC46164.2021.9629513. PMID: 34892695. | Does not meet PICRO (machine learning framework for automatic motion artifact detection on electrodermal activity signals) |
| Jahanbekam A, Baumann J, Nass RD, Bauckhage C, Hill H, Elger CE, Surges R. Performance of ECG-based seizure detection algorithms strongly depends on training and test conditions. Epilepsia Open. 2021 Sep;6(3):597-606. doi: 10.1002/epi4.12520. Epub 2021 Jul 20. PMID: 34250754; PMCID: PMC8408591. | Does not meet PICRO (algorithm) |
| Li C, Zhou W, Liu G, Zhang Y, Geng M, Liu Z, Wang S, Shang W. Seizure Onset Detection Using Empirical Mode Decomposition and Common Spatial Pattern. IEEE Trans Neural Syst Rehabil Eng. 2021;29:458-467. doi: 10.1109/TNSRE.2021.3055276. Epub 2021 Mar 2. PMID: 33507872. | Does not meet PICRO (testing method, no device) |
| Muhammad Usman S, Khalid S, Bashir S. A deep learning based ensemble learning method for epileptic seizure prediction. Comput Biol Med. 2021 Sep;136:104710. doi: 10.1016/j.compbiomed.2021.104710. Epub 2021 Jul 31. PMID: 34364257. | Does not meet PICRO (method to predict epileptic seizures) |
| J.U. Muñoz-Minjares, M. Lopez-Ramirez, Miguel Vazquez-Olguin, C. Lastre-Dominguez, Yuriy S. Shmaliy, Outliers detection for accurate HRV-seizure baseline estimation using modern numerical algorithms, Biomedical Signal Processing and Control, Volume 67, 2021, 102553, ISSN 1746-8094, https://doi.org/10.1016/j.bspc.2021.102553. | Does not meet PICRO (comparison algorithm) |
| Nasseri M, Pal Attia T, Joseph B, Gregg NM, Nurse ES, Viana PF, Schulze-Bonhage A, Dümpelmann M, Worrell G, Freestone DR, Richardson MP, Brinkmann BH. Non-invasive wearable seizure detection using long-short-term memory networks with transfer learning. J Neural Eng. 2021 Apr 8;18(5). doi: 10.1088/1741-2552/abef8a. PMID: 33730713. | Does not meet PICRO (algorithm) |
| Pang TD, Nearing BD, Verrier RL, Schachter SC. T-wave heterogeneity crescendo in the surface EKG is superior to heart rate acceleration for seizure prediction. Epilepsy Behav. 2022 May;130:108670. doi: 10.1016/j.yebeh.2022.108670. Epub 2022 Mar 31. PMID: 35367725. | Does not meet PICRO (prediction seizure onset based on T-wave heterogeneity) |
| Usman SM, Khalid S, Jabbar S, Bashir S. Detection of preictal state in epileptic seizures using ensemble classifier. Epilepsy Res. 2021 Dec;178:106818. doi: 10.1016/j.eplepsyres.2021.106818. Epub 2021 Nov 25. PMID: 34847427. | Does not meet PICRO (no device) |
| Beniczky S, Karoly P, Nurse E, Ryvlin P, Cook M. Machine learning and wearable devices of the future. Epilepsia. 2021 Mar;62 Suppl 2:S116-S124. doi: 10.1111/epi.16555. Epub 2020 Jul 26. PMID: 32712958. | Does not meet PICRO (machine learning) |
| Elezi L, Koren JP, Pirker S, Baumgartner C. Automatic seizure detection and seizure pattern morphology. Clin Neurophysiol. 2022 Jun;138:214-220. doi: 10.1016/j.clinph.2022.02.027. Epub 2022 Mar 18. PMID: 35382982. | double |
| Li C, Zhou W, Liu G, Zhang Y, Geng M, Liu Z, Wang S, Shang W. Seizure Onset Detection Using Empirical Mode Decomposition and Common Spatial Pattern. IEEE Trans Neural Syst Rehabil Eng. 2021;29:458-467. doi: 10.1109/TNSRE.2021.3055276. Epub 2021 Mar 2. PMID: 33507872. | double |
| Behbahani S, Dabanloo NJ, Nasrabadi AM, Dourado A. Prediction of epileptic seizures based on heart rate variability. Technol Health Care. 2016 Nov 14;24(6):795-810. doi: 10.3233/THC-161225. PMID: 27315150. | Wrong study (phase 1 according to classification Benickzy&Ryvlin). |
Verantwoording
Autorisatiedatum en geldigheid
Laatst beoordeeld : 30-05-2023
Laatst geautoriseerd : 28-09-2023
Geplande herbeoordeling : 01-05-2024
Algemene gegevens
De ontwikkeling/herziening van deze richtlijnmodule werd ondersteund door het Kennisinstituut van de Federatie Medisch Specialisten (www.demedischspecialist.nl/kennisinstituut) en werd gefinancierd uit de Kwaliteitsgelden Medisch Specialisten (SKMS). De financier heeft geen enkele invloed gehad op de inhoud van de richtlijnmodule.
Samenstelling werkgroep
Voor het ontwikkelen van de richtlijnmodules is in 2020 een multidisciplinair cluster ingesteld, bestaande uit vertegenwoordigers van alle relevante specialismen die betrokken zijn bij de zorg voor patiënten met epilepsie.
Clusterstuurgroep
- Prof. dr. H.J.M. Majoie (voorzitter), neuroloog, Academisch Centrum voor Epileptologie Kempenhaeghe/ Maastricht UMC+, Heeze en Maastricht
- Drs. M.H.G. Dremmen, radioloog, Erasmus MC Rotterdam
- Dr. P. Klarenbeek, neuroloog, Zuyderland Medisch Centrum, Heerlen
- Dr. J. Nicolai, kinderneuroloog, Academisch Centrum voor Epileptologie Kempenhaeghe/Maastricht UMC+, Maastricht
- Dr. C.M. Delsman-van Gelder, kinderneuroloog, RadboudUMC, Nijmegen
- Dr. P. van Vliet, neuroloog/intensivist, Haaglanden Medisch Centrum, Den Haag
- Drs. R. van Vugt, anesthesioloog, Sint Maartens Kliniek, Nijmegen
Clusterexpertisegroep betrokken bij deze module
- Dr. R.H.C. Lazeron, neuroloog, Academisch Centrum voor Epileptologie Kempenhaeghe/Maastricht UMC+, Heeze
- Dr. R.D. Thijs, neuroloog, SEIN, Heemstede en LUMC, Leiden
- Dr. A. Uiterwijk, neuroloog, Academisch Centrum voor Epileptologie, Kempenhaeghe/Maastricht UMC+, Heeze
Met ondersteuning van
- Dr. M.M.J. van Rooijen, adviseur Kennisinstituut van de Federatie Medisch Specialisten Utrecht
- Dr. J. Buddeke senior adviseur Kennisinstituut van de Federatie Medisch Specialisten Utrecht
Belangenverklaringen
De Code ter voorkoming van oneigenlijke beïnvloeding door belangenverstrengeling is gevolgd. Alle clusterleden hebben schriftelijk verklaard of zij in de laatste drie jaar directe financiële belangen (betrekking bij een commercieel bedrijf, persoonlijke financiële belangen, onderzoeksfinanciering) of indirecte belangen (persoonlijke relaties, reputatiemanagement) hebben gehad. Gedurende de ontwikkeling of herziening van een module worden wijzigingen in belangen aan de voorzitter doorgegeven. De belangenverklaring wordt opnieuw bevestigd tijdens de commentaarfase.
Een overzicht van de belangen van de clusterleden en het oordeel over het omgaan met eventuele belangen vindt u in onderstaande tabel. De ondertekende belangenverklaringen zijn op te vragen bij het secretariaat van het Kennisinstituut van de Federatie Medisch Specialisten.
Stuurgroep
| Clusterlid | Functie | Nevenfuncties | Gemelde belangen | Ondernomen actie |
| Majoie* | Functie: Neurologie Werkgever: Academisch Centrum voor Epileptologie Kempenhaeghe, Maastricht UMC+ | Relevante commissies
NB geen betaalde functies | Lopende onderzoek- en zorginnovatieprojecten (anders dan contract research) worden gefinancierd uit ZonMW, Nationaal epilepsiefonds, stichting vrienden van Kempenhaeghe, en SKMS.
Incidenteel financiële ondersteuning aan stichting Kempenhaeghe voor organisatie refereeravonden en workshops/symposia (telkens volgens geldende wet en regelgeving). | Geen actie |
| Delsman | Kinderneuroloog/kinderarts Werkgever: Maxima Medisch Centrum Veldhoven | Geen | Geen | Geen actie |
| Dremmen | Functie: Kinderradioloog met subspecialisatie kinderneuroradiologie Werkgever: ErasmusMC te Rotterdam | Een van de voorzitters van de richtlijn Radiologische diagnostiek acute trauma-opvang bij kinderen Expertisegroep richtlijn Traumatologie Uitgenodigd spreker op meerdere nationale en internationale cursussen en congressen | Geen | Geen actie |
| Klarenbeek | Functie: Neuroloog Werkgever: Zuyderland te Heerlen/Sittard (vrijgevestigd) | Geen | Geen | Geen actie |
| Nicolai | Functie: Kinderneuroloog Werkgever: Maastricht UMC+. Tevens gedetacheerd/ werkzaam in Kempenhaeghe, Heeze; St Jansgasthuis, Weert; Elkerliek, Helmond en Viecuri, Venlo. |
Coördinator Sepion beginnerscursus; vergoeding wordt door Sepion naar AMBI goed doel gestort. Tevens betrokken bij Sepion gevorderden cursus en AVG cursus; vergoeding wordt door Sepion naar AMBI goed doel gestort. | Boegbeeldfunctie bij een patiënten- of beroepsorganisatie | Geen actie |
| Van Vliet | Functie: Intensivist Werkgever: Haaglanden Medisch Centrum te Den Haag |
| Geen | Geen actie |
| Van Vugt | Functie: Anesthesioloog Werkgever: Sint Maartenskliniek te Nijmegen | Lid commissie kwaliteitsdocumenten NVA (onbetaald) | Geen | Geen actie |
Expertisegroep
| Clusterlid | Functie | Nevenfuncties | Gemelde belangen | Ondernomen actie |
| Altinbas | Functie: Neuroloog-kinderneuroloog Werkgever: SEIN locatie Meer en Bosch werkzaam op de polikliniek en kliniek (0.6 fte), LUMC polikliniek kinderneurologie (0.2 fte). |
| Per 2023 zal het landelijk CBD consortium een onderzoek starten naar de behandeling van kinderen met therapieresistente epilepsie met CBD (gesubsidieerd door ZonMW). | Geen actie |
| Balvers | Functie: Neuroloog (Behandeling van patiënten met epilepsie; klinisch, poliklinisch en in woonzorg) Werkgever: Stichting Epilepsie Instellingen Nederland (SEIN) (0.6fte)
Functie: Neuroloog (Behandeling van patiënten met hoofdpijn; poliklinisch) Werkgever: The Migraine Clinic, Amsterdam (0.2 fte) |
|
| Geen restrictie voor Epilepsie. |
| De Bruijn | Functie: neuroloog in opleiding Werkgever: ETZ Tilburg | Geen | Geen | Geen actie |
| Draak | Functie: Kinderneuroloog, neuroloog. Werkgever: Zuyderland Medisch Centrum | Geen | Geen | Geen actie |
| Eshuis | Functie: AIOS Spoedeisende geneeskunde Werkgever: Catharina Ziekenhuis te Eindhoven |
| Geen | Geen actie |
| Hofman | Functie: Radioloog Werkgever: Maastricht UMC+ (1.0fte) |
| Geen | Geen actie |
| Koekkoek | Functie: Neuroloog Werkgever: Leids Universitair Medisch Centrum (0.8fte), Haaglanden Medisch Centrum (0.2fte) |
| Geen | Geen actie |
| Lazeron | Functie: Neuroloog Werkgever: Academisch centrum Epileptologie Kempenhaeghe Maastricht UMC+ (voltijds).
Functie: Wetenschappelijk onderzoeker Werkgever: Technische universiteit te Eindhoven (nul uren aanstelling) |
| Lid van het TeleEpilepsie consortium. Met dit consortium hebben een aanvalsdetector ontwikkeld, de Nightwatch(R) die door LivAssured geproduceerd wordt. Geen direct financieel belang in dit bedrijf, maar van eventuele toekomstige winsten van LivAssured vloeit een klein percentage terug naar het consortium voor epilepsie-onderzoek | Restricties ten aanzien van besluitvorming met betrekking tot apparaten die aanvallen detecteren (e.g. Nightwatch) |
| Masselink | Functie: Ziekenhuisapotheker Werkgever: Medisch Spectrum Twente |
| Geen | Geen actie |
| M’Rabet | Functie: MT-lid Kennis en Innovatie Werkgever: EpilepsieNL, Houten | Geen |
| Geen actie |
| Reijneveld | Functie: Neuroloog, Werkgever: Stichting Epilepsie Instellingen Nederland (SEIN) te Heemstede (0.8 fte)
Functie: Universitair Hoofddocent, afdeling Neurologie Werkgever: Amsterdam UMC (0.2 fte) |
| Geen | Geen actie |
| Ronner | Functie: Neuroloog Werkgever: Amsterdam UMC (0.8 fte) |
| Geen | Geen actie |
| Schijns | Functie: Neurochirurg met specialisatie Epilepsiechirurgie en Neuro-oncologische Chirurgie Werkgever: Maastricht UMC+ |
|
| Geen actie |
| Snoeijen | Functie: Arts Verstandelijk Gehandicapten Werkgever: Kempenhaeghe, fulltime | Geen | Geen | Geen actie |
| Thijs | Functie: neuroloog Werkgever: Stichting Epilepsie Instellingen Nederland (1.0 fte) Leids Universitair Medisch Centrum (detachering vanuit SEIN voor 0.2 fte) |
|
| Restrictie ten aanzien van besluitvorming met betrekking tot apparaten die aanvallen detecteren (e.g. Nightwatch®). |
| Tousseyn | Functie: neuroloog Werkgever: Academisch Centrum voor Epileptologie (ACE) Kempenhaeghe/Maastricht UMC+, locatie Heeze |
|
| Geen actie |
| Uiterwijk | Functie: Epileptoloog Werkgever: Academisch Centrum voor Epileptologie Kempenhaeghe | Geen | Geen | Geen actie |
| Van 't Hof | Functie: neuroloog Werkgever: SEIN (Stichting Epilepsie Instellingen Nederland) |
| Geen | Geen actie |
| Van Tuijl | Functie: Neuroloog Werkgever: ETZ Tilburg |
| Geen actie | |
| Tolboom | Functie: Nucleair geneeskundige Werkgever: UMC Utrecht | Geen | Geen | Geen actie |
| Verbeek | Functie: Klinisch geneticus Werkgever: UMC Utrecht | Geen | Geen | Geen actie |
| Vlooswijk | Functie: Neuroloog Werkgever: Maastricht UMC+ |
|
| Geen actie |
| Wegner | Functie: Neuroloog Werkgever: Stichting Epilepsie instellingen Nederland (SEIN) |
| Geen | Geen actie |
In de ontwikkelfase van de module Aanvalsdetectie zijn de conceptaanbevelingen plenair tijdens de werkgroepvergadering door de gehele werkgroep geformuleerd. In de commentaarfase zijn onafhankelijke reviewers aangesteld om de conceptmodule van commentaar te voorzien.
Inbreng patiëntenperspectief
Er werd aandacht besteed aan het patiëntenperspectief door de afvaardiging van EpilepsieNL in het cluster. De verkregen input is meegenomen bij het opstellen van de uitgangsvragen, de keuze voor de uitkomstmaten en bij het opstellen van de overwegingen. De conceptmodule is tevens voor commentaar voorgelegd aan Epilepsie NL en de eventueel aangeleverde commentaren zijn bekeken en verwerkt.
Kwalitatieve raming van mogelijke financiële gevolgen in het kader van de Wkkgz
Bij de richtlijn is conform de Wet kwaliteit, klachten en geschillen zorg (Wkkgz) een kwalitatieve raming uitgevoerd of de aanbevelingen mogelijk leiden tot substantiële financiële gevolgen. Bij het uitvoeren van deze beoordeling zijn richtlijnmodules op verschillende domeinen getoetst (zie het stroomschema op de Richtlijnendatabase).
Uit de kwalitatieve raming blijkt dat er [waarschijnlijk geen/ mogelijk] substantiële financiële gevolgen zijn, zie onderstaande tabel.
| Module | Uitkomst raming | Toelichting |
| Module ‘Aanvalsdetectie’ | geen financiële gevolgen |
Werkwijze
AGREE
Deze richtlijnmodule is opgesteld conform de eisen vermeld in het rapport Medisch Specialistische Richtlijnen 2.0 van de adviescommissie Richtlijnen van de Raad Kwaliteit. Dit rapport is gebaseerd op het AGREE II instrument (Appraisal of Guidelines for Research & Evaluation II; Brouwers, 2010).
Cyclus werkwijze
Bij de start van de cyclus is de geldigheid van alle modules geïnventariseerd, waarna er een prioriteringsronde heeft plaatsgevonden. De geprioriteerde modules zijn herzien en geüpdatet door het cluster.
Uitkomstmaten
Na het opstellen van de zoekvraag behorende bij de uitgangsvraag inventariseerde het cluster welke uitkomstmaten voor de patiënt relevant zijn, waarbij zowel naar gewenste als ongewenste effecten werd gekeken. Hierbij werd een maximum van acht uitkomstmaten gehanteerd. Het cluster waardeerde deze uitkomstmaten volgens hun relatieve belang bij de besluitvorming rondom aanbevelingen, als cruciaal (kritiek voor de besluitvorming), belangrijk (maar niet cruciaal) en onbelangrijk. Tevens definieerde het cluster tenminste voor de cruciale uitkomstmaten welke verschillen zij klinisch (patiënt) relevant vonden.
Methode literatuursamenvatting
Een uitgebreide beschrijving van de strategie voor zoeken en selecteren van literatuur is te vinden onder ‘Zoeken en selecteren’ onder Onderbouwing. Indien mogelijk werd de data uit verschillende studies gepoold in een random-effects model. Review Manager 5.4 werd gebruikt voor de statistische analyses. De beoordeling van de kracht van het wetenschappelijke bewijs wordt hieronder toegelicht.
Beoordelen van de kracht van het wetenschappelijke bewijs
De kracht van het wetenschappelijke bewijs werd bepaald volgens de GRADE-methode. GRADE staat voor ‘Grading Recommendations Assessment, Development and Evaluation’ (zie http://www.gradeworkinggroup.org/). De basisprincipes van de GRADE-methodiek zijn: het benoemen en prioriteren van de klinisch (patiënt) relevante uitkomstmaten, een systematische review per uitkomstmaat, en een beoordeling van de bewijskracht per uitkomstmaat op basis van de acht GRADE-domeinen (domeinen voor downgraden: risk of bias, inconsistentie, indirectheid, imprecisie, en publicatiebias; domeinen voor upgraden: dosis-effect relatie, groot effect, en residuele plausibele confounding).
GRADE onderscheidt vier gradaties voor de kwaliteit van het wetenschappelijk bewijs: hoog, redelijk, laag en zeer laag. Deze gradaties verwijzen naar de mate van zekerheid die er bestaat over de literatuurconclusie, in het bijzonder de mate van zekerheid dat de literatuurconclusie de aanbeveling adequaat ondersteunt (Schünemann, 2013; Hultcrantz, 2017).
| GRADE | Definitie |
| Hoog |
|
| Redelijk |
|
| Laag |
|
| Zeer laag |
|
Bij het beoordelen (graderen) van de kracht van het wetenschappelijk bewijs in richtlijnen volgens de GRADE-methodiek spelen grenzen voor klinische besluitvorming een belangrijke rol (Hultcrantz, 2017). Dit zijn de grenzen die bij overschrijding aanleiding zouden geven tot een aanpassing van de aanbeveling. Om de grenzen voor klinische besluitvorming te bepalen moeten alle relevante uitkomstmaten en overwegingen worden meegewogen. De grenzen voor klinische besluitvorming zijn daarmee niet één op één vergelijkbaar met het minimaal klinisch relevant verschil (Minimal Clinically Important Difference, MCID). Met name in situaties waarin een interventie geen belangrijke nadelen heeft en de kosten relatief laag zijn, kan de grens voor klinische besluitvorming met betrekking tot de effectiviteit van de interventie bij een lagere waarde (dichter bij het nuleffect) liggen dan de MCID (Hultcrantz, 2017).
Netwerk meta-analyse
Voor de beoordeling van de bewijskracht uit een netwerk meta-analyse (NMA) wordt gebruik gemaakt van de CINeMA-tool (Nikolakopoulou, 2020; Papakonstantinou, 2020). Met deze tool wordt op basis van een analyse in R de effectschattingen berekend. Voor het beoordelen van de bewijskracht, op basis van de berekende effectschattingen, worden zes domeinen geëvalueerd, namelijk Risk-of-Bias, publicatiebias, indirectheid, imprecisie, heterogeniteit and incoherentie. Voor het opstellen van de conclusies zijn principes van de GRADE working group gebruikt op basis van grenzen van klinische besluitvorming (Brignardello-Petersen, 2020).
Overwegingen (van bewijs naar aanbeveling)
Om te komen tot een aanbeveling zijn naast (de kwaliteit van) het wetenschappelijke bewijs ook andere aspecten belangrijk en worden meegewogen, zoals aanvullende argumenten uit bijvoorbeeld de biomechanica of fysiologie, waarden en voorkeuren van patiënten, kosten (middelenbeslag), aanvaardbaarheid, haalbaarheid en implementatie. Deze aspecten zijn systematisch vermeld en beoordeeld (gewogen) onder het kopje ‘Overwegingen’ en kunnen (mede) gebaseerd zijn op expert opinion. Hierbij is gebruik gemaakt van een gestructureerd format gebaseerd op het evidence-to-decision framework van de internationale GRADE Working Group (Alonso-Coello, 2016a; Alonso-Coello 2016b). Dit evidence-to-decision framework is een integraal onderdeel van de GRADE methodiek.
Formuleren van aanbevelingen
De aanbevelingen geven antwoord op de uitgangsvraag en zijn gebaseerd op het beschikbare wetenschappelijke bewijs en de belangrijkste overwegingen, en een weging van de gunstige en ongunstige effecten van de relevante interventies. De kracht van het wetenschappelijk bewijs en het gewicht dat door het cluster wordt toegekend aan de overwegingen, bepalen samen de sterkte van de aanbeveling. Conform de GRADE-methodiek sluit een lage bewijskracht van conclusies in de systematische literatuuranalyse een sterke aanbeveling niet a priori uit, en zijn bij een hoge bewijskracht ook zwakke aanbevelingen mogelijk (Agoritsas, 2017; Neumann, 2016). De sterkte van de aanbeveling wordt altijd bepaald door weging van alle relevante argumenten tezamen. Het cluster heeft bij elke aanbeveling opgenomen hoe zij tot de richting en sterkte van de aanbeveling zijn gekomen.
In de GRADE-methodiek wordt onderscheid gemaakt tussen sterke en zwakke (of conditionele) aanbevelingen. De sterkte van een aanbeveling verwijst naar de mate van zekerheid dat de voordelen van de interventie opwegen tegen de nadelen (of vice versa), gezien over het hele spectrum van patiënten waarvoor de aanbeveling is bedoeld. De sterkte van een aanbeveling heeft duidelijke implicaties voor patiënten, behandelaars en beleidsmakers (zie onderstaande tabel). Een aanbeveling is geen dictaat, zelfs een sterke aanbeveling gebaseerd op bewijs van hoge kwaliteit (GRADE gradering HOOG) zal niet altijd van toepassing zijn, onder alle mogelijke omstandigheden en voor elke individuele patiënt.
| Implicaties van sterke en zwakke aanbevelingen voor verschillende richtlijngebruikers | ||
| Sterke aanbeveling | Zwakke (conditionele) aanbeveling | |
| Voor patiënten | De meeste patiënten zouden de aanbevolen interventie of aanpak kiezen en slechts een klein aantal niet. | Een aanzienlijk deel van de patiënten zouden de aanbevolen interventie of aanpak kiezen, maar veel patiënten ook niet. |
| Voor behandelaars | De meeste patiënten zouden de aanbevolen interventie of aanpak moeten ontvangen. | Er zijn meerdere geschikte interventies of aanpakken. De patiënt moet worden ondersteund bij de keuze voor de interventie of aanpak die het beste aansluit bij zijn of haar waarden en voorkeuren. |
| Voor beleidsmakers | De aanbevolen interventie of aanpak kan worden gezien als standaardbeleid. | Beleidsbepaling vereist uitvoerige discussie met betrokkenheid van veel stakeholders. Er is een grotere kans op lokale beleidsverschillen. |
Organisatie van zorg
Bij de ontwikkeling van de richtlijnmodule is expliciet aandacht geweest voor de organisatie van zorg: alle aspecten die randvoorwaardelijk zijn voor het verlenen van zorg (zoals coördinatie, communicatie, (financiële) middelen, mankracht en infrastructuur). Randvoorwaarden die relevant zijn voor het beantwoorden van deze specifieke uitgangsvraag zijn genoemd bij de overwegingen. Meer algemene, overkoepelende, of bijkomende aspecten van de organisatie van zorg worden behandeld in de module Organisatie van zorg.
Commentaar- en autorisatiefase
De conceptrichtlijnmodule werd aan de betrokken (wetenschappelijke) verenigingen en (patiënt) organisaties voorgelegd ter commentaar. De commentaren werden verzameld en besproken met het cluster. Naar aanleiding van de commentaren werd de conceptrichtlijnmodule aangepast en definitief vastgesteld door het cluster. De definitieve richtlijnmodule werd aan de deelnemende (wetenschappelijke) verenigingen en (patiënt)organisaties voorgelegd voor autorisatie en door hen geautoriseerd dan wel geaccordeerd.
Literatuur
Agoritsas T, Merglen A, Heen AF, Kristiansen A, Neumann I, Brito JP, Brignardello-Petersen R, Alexander PE, Rind DM, Vandvik PO, Guyatt GH. UpToDate adherence to GRADE criteria for strong recommendations: an analytical survey. BMJ Open. 2017 Nov 16;7(11):e018593. doi: 10.1136/bmjopen-2017-018593. PubMed PMID: 29150475; PubMed Central PMCID: PMC5701989.
Alonso-Coello P, Schünemann HJ, Moberg J, Brignardello-Petersen R, Akl EA, Davoli M, Treweek S, Mustafa RA, Rada G, Rosenbaum S, Morelli A, Guyatt GH, Oxman AD; GRADE Working Group. GRADE Evidence to Decision (EtD) frameworks: a systematic and transparent approach to making well informed healthcare choices. 1: Introduction. BMJ. 2016 Jun 28;353:i2016. doi: 10.1136/bmj.i2016. PubMed PMID: 27353417.
Alonso-Coello P, Oxman AD, Moberg J, Brignardello-Petersen R, Akl EA, Davoli M, Treweek S, Mustafa RA, Vandvik PO, Meerpohl J, Guyatt GH, Schünemann HJ; GRADE Working Group. GRADE Evidence to Decision (EtD) frameworks: a systematic and transparent approach to making well informed healthcare choices. 2: Clinical practice guidelines. BMJ. 2016 Jun 30;353:i2089. doi: 10.1136/bmj.i2089. PubMed PMID: 27365494.
Brignardello-Petersen R, Florez ID, Izcovich A, Santesso N, Hazlewood G, Alhazanni W, Yepes-Nuñez JJ, Tomlinson G, Schünemann HJ, Guyatt GH; GRADE working group. GRADE approach to drawing conclusions from a network meta-analysis using a minimally contextualised framework. BMJ. 2020 Nov 11;371:m3900. doi: 10.1136/bmj.m3900. PMID: 33177059.
Brouwers MC, Kho ME, Browman GP, Burgers JS, Cluzeau F, Feder G, Fervers B, Graham ID, Grimshaw J, Hanna SE, Littlejohns P, Makarski J, Zitzelsberger L; AGREE Next Steps Consortium. AGREE II: advancing guideline development, reporting and evaluation in health care. CMAJ. 2010 Dec 14;182(18):E839-42. doi: 10.1503/cmaj.090449. Epub 2010 Jul 5. Review. PubMed PMID: 20603348; PubMed Central PMCID: PMC3001530.
Hultcrantz M, Rind D, Akl EA, Treweek S, Mustafa RA, Iorio A, Alper BS, Meerpohl JJ, Murad MH, Ansari MT, Katikireddi SV, Östlund P, Tranæus S, Christensen R, Gartlehner G, Brozek J, Izcovich A, Schünemann H, Guyatt G. The GRADE Working Group clarifies the construct of certainty of evidence. J Clin Epidemiol. 2017 Jul;87:4-13. doi: 10.1016/j.jclinepi.2017.05.006. Epub 2017 May 18. PubMed PMID: 28529184; PubMed Central PMCID: PMC6542664.
Medisch Specialistische Richtlijnen 2.0 (2012). Adviescommissie Richtlijnen van de Raad Kwalitieit. http://richtlijnendatabase.nl/over_deze_site/over_richtlijnontwikkeling.html
Neumann I, Santesso N, Akl EA, Rind DM, Vandvik PO, Alonso-Coello P, Agoritsas T, Mustafa RA, Alexander PE, Schünemann H, Guyatt GH. A guide for health professionals to interpret and use recommendations in guidelines developed with the GRADE approach. J Clin Epidemiol. 2016 Apr;72:45-55. doi: 10.1016/j.jclinepi.2015.11.017. Epub 2016 Jan 6. Review. PubMed PMID: 26772609.
Nikolakopoulou A, Higgins JPT, Papakonstantinou T, Chaimani A, Del Giovane C, Egger M, Salanti G. CINeMA: An approach for assessing confidence in the results of a network meta-analysis. PLoS Med. 2020 Apr 3;17(4):e1003082. doi: 10.1371/journal.pmed.1003082. PMID: 32243458; PMCID: PMC7122720.
Schünemann H, Brożek J, Guyatt G, et al. GRADE handbook for grading quality of evidence and strength of recommendations. Updated October 2013. The GRADE Working Group, 2013. Available from http://gdt.guidelinedevelopment.org/central_prod/_design/client/handbook/handbook.html.
Papakonstantinou T, Nikolakopoulou A, Higgins JPT, Egger M & Salanti G. CINeMA: Software for semiautomated assessment of the confidence in the results of network meta-analysis. Campbell Systematic Reviews 2020;16:e1080.
Zoekverantwoording
Zoekacties zijn opvraagbaar. Neem hiervoor contact op met de Richtlijnendatabase.
Bijlagen
- Achtergrond en definities
- Toepassen
- Overige