Health‐related quality of life in narcolepsy: A systematic review and meta‐analysis

1 INTRODUCTION

Narcolepsy is a rare, disabling chronic neurological disorder that is characterised by excessive daytime sleepiness (EDS), cataplexy, hypnagogic hallucinations and sleep paralysis. Narcolepsy can be classified into two subtypes: type 1 narcolepsy (NT1) and type 2 narcolepsy (NT2), both of which have similar clinical presentations. However, NT1 can be distinguished by the presence of cataplexy, which is defined as an episodic loss of muscle tone in full consciousness that typically arises following intense emotions such as laughter or anger and decreased cerebrospinal fluid levels of hypocretin (Sateia, 2014). The incidence of narcolepsy is estimated to be 25–50 per 100,000 in Western populations (Overeem, Black, & Lammers, 2008). Symptom onset typically occurs in adolescence; however, approximately one-third of people with narcolepsy experience initial symptoms in adulthood (Dauvilliers et al., 2001).

Health-related quality of life (HRQoL) can be described as a “multidimensional concept that includes subjective reports of symptoms, side effects, functioning in multiple life domains, and general perceptions of life satisfaction and quality” (Revicki, Kleinman, & Cella, 2014). Narcolepsy is a neurological condition that can predispose to the development of social and occupational dysfunction (Morse & Sanjeev, 2018). This condition has been associated with considerable detriment to daily life, including impaired quality of life, occupational and academic difficulties, and adversely affected social and personal relationships (Emsellem et al., 2020; Flores, Villa, Black, Chervin, & Witt, 2016; Kapella et al., 2015). With significant correlations identified between symptom severity and HRQoL (Dauvilliers et al., 2017), mitigating the deleterious effect of narcolepsy on HRQoL should be a critical therapeutic goal for people with narcolepsy.

Despite the frequent inclusion of HRQoL as an outcome measure in narcolepsy trials; to date, there has been no systematic review and meta-analysis to synthesise the literature and provide a summative assessment of the impact of narcolepsy on HRQoL. The aim of this review was to systematically review the literature assessing HRQoL in people with narcolepsy, provide pooled mean scores of the domains of the various HRQoL tools used in this population if possible, and to compare HRQoL in people with narcolepsy with general population norms and other chronic health conditions. Additional objectives of this review are to explore: (a) the heterogeneity of the published studies; (b) the tools used to assess HRQoL in this population; and (c) the influence of study characteristics on HRQoL.

2 METHODS

This systematic review sought to identify the HRQoL of people with narcolepsy. This review followed the “Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)” statement guidelines. A study protocol that included the proposed search strategy and methodology was registered with PROSPERO, the international prospective registry of systematic reviews (PROSPERO) database (www.crd.york.ac.uk/prospero/) in April 2020 (Identification number: CRD42020156036).

2.1 Eligibility criteria

The target population for this review was people with narcolepsy recruited from the general population, primary care or secondary care settings. Observational studies (case–control, cohort and cross-sectional) and experimental studies (randomised control trials, pre-post design, quasi-experimental) were deemed eligible if they assessed HRQoL in people with narcolepsy using a validated HRQoL questionnaire. HRQoL has been defined as “a term referring to the health aspects of quality of life, generally considered to reflect the impact of disease and treatment on disability and daily functioning; it has also been considered to reflect the impact of perceived health on an individual's ability to live a fulfilling life. However, more specifically HRQoL is a measure of the value assigned to duration of life as modified by impairments, functional states, perceptions and opportunities, as influenced by disease, injury, treatment and policy” (Ahmed & Andrich, 2015). Articles were deemed ineligible for inclusion if they were case-series, case reports, expert opinion or consensus statements, or deduplicate studies that utilised the same participant data. Studies were required to provide mean scores with standard deviations (SDs) or standard errors (SEs) for each domain of their chosen HRQoL tool to be eligible for inclusion for each respective meta-analysis. Articles were restricted to those published in English; however, no limitation was placed on the publication year of articles.

2.2 Data sources and search strategy

In collaboration with a senior medical librarian with specialist skills in systematic review searching (DM), a comprehensive search strategy was developed. The search encompassed four electronic databases: CINAHL, EMBASE, Medline (OVID) and Web of Science. The terms searched consisted of keywords and subject headings that were adapted for each database, and can be divided into three categories: (a) the condition (e.g. “narcolepsy”, “narcolepsy type 1”, “narcolepsy type 2”, “narcolepsy with cataplexy”); (b) HRQoL (“quality of life”, “quality of life assessment”, “HRQoL”); and (c) HRQoL tools (e.g. “Short Form 36”, “European Quality of Life 5 Dimensions Visual Analogue Scale”, “functional outcome of sleep questionnaire”). The reference lists of articles identified in the initial search were scanned to identify any studies potentially missed.

2.3 Selection of eligible studies

Articles were retrieved and deduplicated. Titles and abstracts were screened to determine their eligibility for inclusion by two researchers (RT and JB). Inter-rater disagreements were resolved through careful re-examination and discussion of the article between reviewers until a consensus was reached. The full texts of the potentially eligible studies were retrieved and independently assessed by both reviewers (RT and JB) to determine eligibility for inclusion in the final analyses. A similar method of addressing disagreements between researchers was applied for the full-text screening phase.

2.4 Data extraction and quality assessment

Primary data extraction was conducted by RT, with JB examining the articles independently to reduce bias. Two researchers (RT and JB) independently appraised the risk of bias of included studies, with disagreements resolved through discussion between researchers until a conclusion was reached. A modified version of the Joanna Briggs Institute Checklist for Analytical Cross-Sectional Studies (Moola et al., 2017) was utilised to assess the risk of bias of included studies. This modified tool utilised the following five domains to assess bias: (a) sample; (b) subjects and setting; (c) objective measures of disease; (d) outcome measured; and (e) statistical analysis. Any discrepancies were resolved through discussion and review of the original article. If included articles were longitudinal or follow-up studies, baseline HRQoL data were selected for analysis.

2.5 Statistical analysis

Statistical heterogeneity was determined using I2-values, with values nearing 25%, 50% and 75% representing low, moderate and high heterogeneity, respectively (Higgins, Thompson, Deeks, & Altman, 2003). As high levels of heterogeneity were identified between studies, random-effects meta-analyses with 95% confidence intervals (CIs) using Comprehensive Meta-Analysis were employed. Meta-analyses were conducted for each domain of the Short Form 36 (SF36), and the utility and visual analogue scale (VAS) scores of the EQ5D. Two separate meta-analyses were conducted for the physical (PCS) and mental (MCS) component summaries for the SF36, respectively. The first meta-analyses included only studies that provided calculated PCS and MCS values and their SDs. The second meta-analysis utilised the formula outlined by Taft, Karlsson, and Sullivan (2001) to calculate the PCS and MCS values from the domain scores when summary scores were not provided. SDs for these PCS and MCS scores were imputed according to the process outlined by Furukawa, Barbui, Cipriani, Brambilla, and Watanabe (2006). HRQoL questionnaires that were unable to be meta-analysed were discussed in a narrative summary. The impact of study variables and characteristics on HRQoL was assessed using Spearman's correlation analyses with adjusted r2. The HRQoL of people with narcolepsy was compared against normative SF36 values obtained from the USA (Ware, Snow, Kosinski, & Gandek, 1993), UK (Jenkinson, Coulter, & Wright, 1993), France (Audureau, Rican, & Coste, 2013) and Norway (Ribu, Hanestad, Moum, Birkeland, & Rustoen, 2007). Data from people with narcolepsy were also plotted alongside data from people with epilepsy (Hermann et al., 1996), multiple sclerosis (Hermann et al., 1996), diabetes (Ribu et al., 2007) and hypertension (Kusek et al., 2002).

3 RESULTS 3.1 Study screening

The search strategy yielded 5,706 articles and, following deduplication, 3,399 unique articles had their titles and abstracts assessed for eligibility. From these articles, 3,338 articles were deemed ineligible and excluded. The full texts of the remaining 61 articles were screened to determine eligibility for inclusion, and 31 were excluded; with 24 being published abstracts, three duplicate data sets, three utilising ineligible outcome measures, and one study that assessed people without a formal diagnosis of narcolepsy. The remaining 30 articles were included in a descriptive synthesis, of which 17 articles were included in the SF36 meta-analysis, and five in the EQ5D meta-analysis. Figure 1 shows the study selection process.

image

Flow diagram of studies screened for eligibility

3.2 Characteristics of included studies

The characteristics of the included studies are outlined in Table 1. The 30 reviewed studies represent a total sample of 4,600 people with narcolepsy, of which 54.31% were female (n = 2,498). The number of participants in each study ranged from 15 to 558, with a mean of 153 participants in the included studies. The mean age of all participants was 40.8 years, with a 95% CI ranging from 37.12 to 44.46 years. The 30 included studies originated from 13 different countries (Table 1). Studies were predominantly based in North America and Europe (80.00%), and approximately one-third of studies (n = 8) were published in the USA (Becker, Schwartz, Feldman, & Hughes, 2004; Beusterien et al., 1999; Bogan et al., 2017; Emsellem et al., 2020; Flores et al., 2016; Kapella et al., 2015; Mitler, Harsh, Hirshkowitz, & Guilleminault, 2000; Weaver & Cuellar, 2006). Four studies were published in France (Dauvilliers et al., ,2009, 2011, 2017, 2019) and Italy (Ingravallo et al., ,2008, 2012; Vignatelli et al., 2004; Vignatelli, Plazzi, Peschechera, Delaj, & D’Alessandro, 2011), respectively. Additionally, three studies were published in Japan (Kayaba, Sasai -Sakuma, & Inoue, 2018; Ozaki et al., ,2008, 2012). One study was published from each of the remaining countries (Table 1).

TABLE 1. Study characteristics Author Year Country Study type Industry funding Sample size M/F Age (years) Instrument Quality Score Comparison Groups/Control Becker 2004 America Cohort study Yes 151 70/81 39.00 (18–68) SF36 4/5 No control group

Beusterien

1999 America RCT NR 481 251/307 42 SF36 with additional scales 3/5 Placebo control Bogan 2017 America Post hoc analysis Yes 228 79/149 40.50 (± 15.30) SF-36 4/5 Placebo control Campell 2011 New Zealand Cross-sectional No 54 20/34 54.70 (± 18.30) SF36 3/5 No control group Daniels 2001 UK Cross-sectional No 305 120/185 56.00 (18–89) SF36 3/5 No control group Dauvilliers 2009 France Cross-sectional Yes 492 238/254 41.64 (± 16.53) SF36 5/5 No Control Group: Compared NT1, NT2 and IH Dauvilliers 2011 France Cross-sectional Yes 67 31/36

44.8% < 40;

55.2% > 40 

SF36, FOSQ 5/5 Compared with matched controls Dauvilliers 2017 France Cross-sectional NR 175 104/71 41.50 (± 17.36) EQ5D 4/5 Compared drug-free and treated patients Dauvilliers 2019 France Cross-sectional NR 39 22/17 39.45 (± 18.20) EQ5D 4/5 No control group: compared IH with NT1 David 2012 Portugal Cross-sectional NR 51 26/25 43.40 (± 15.30) SF36 4/5 Compared with population norms Dodel 2007 Germany Cross-sectional NR 75 46/29 48.90 (± 15.20) SF36, EQ5D 4/5 Compared with population norms Droogleever Fortuyn 2012 Netherlands Cross-sectional NR 80 46/34 48.3 (± 14.70) SF36 3/5 No Control Group Compared fatigued versus non-fatigued Emsellem 2020 America RCT Yes 231 82/154 36.23 (± 13.20) SF36, EQ5D, FOSQ 4/5 Placebo control Ervik 2006 Norway Cross-sectional NR 77 16/54 53.0 (± 17.40) SF36 5/5 No control group Flores 2016 America Cross-sectional Yes 437 219/218 46.70 (± 16.40) SF36 PCS and MCS 3/5 Compared with population norms Ingravallo 2008 Italy Cross-sectional No 15 9/6 48.70 (± 18.80) SF36 PCS and MCS 5/5 No control group Ingravallo 2012 Italy Cross-sectional NR 100 51/49 37.10 (18–65) EQ5D 3/5 Compared with population norms Kapella 2015 America Cross-sectional No 122 27/95 27.10 (± 5.00) SF36, FOSQ 3/5 Acquaintance Approach for control group Kayaba 2018 Japan Cross-sectional No 39 20/119 24.60 (± 8.30) SF36 PCS and MCS 4/5 Compared with BIISS and DSPD Kovalska 2016 Czech Republic Case–control No 42 18/24 71.86 (± 7.45) VAS EQ5D 5/5 Age- and gender-matched controls Mitler 2000 America Cross-sectional No 478 220/258 42.00 (± 13.0) SF36 4/5 No control group Ozaki 2008 Japan Cross-sectional No 55 20/35 30.29 (± 10.59) SF36 5/5 Treated versus drug-naïve Ozaki 2012 Japan Cross-sectional No 131 71/63 32.21 (± 8.68) SF36 5/5 Treated versus drug-naïve Rovere 2008 Brazil Cross-sectional No 40 12/28 41.85 (± 14.5) WHOQOL-BREF 3/5 Control group present Sarkanen 2016 Finland Cross-sectional NR 51 25/26 NR WHO-5 Well-Being Index 4/5 Compared with NT1 Song 2019 South Korea Cross-sectional No 63 43/20 27.03 (± 9.29) K-SF36 5/5 No control group Teixeira 2004 Scotland Cross-sectional No 49 30/19 47.00 (± 18.00) SF36, FOSQ 5/5 Untreated OSAHS and CPAP-treated OSAHS Vignatelli 2004 Italy Cross-sectional No 108 62/46 43.20 (± 16.40) SF36 5/5 Compared with population norms Vignatelli 2011 Italy 5-year prospective cohort No 54 42/12 48.00 (± 18.40) SF36 5/5 5-year follow-up Weaver 2006 America RCT Yes 228 79/149 40.50 (± 15.30) FOSQ 5/5 Placebo control BISS, behaviourally induced insufficient sleep syndrome; CPAP, continuous positive airway pressure; DSPD, delayed sleep phase disorder; FOSQ, Functional Outcome of Sleep Questionnaire; IH, idiopathic hypersomnia; K-SF36, Korean Short Form 36; MCS, mental component summary; NR, not reported; NT1, type 1 narcolepsy; NT2, type 2 narcolepsy; OSAHS, obstructive sleep apnea hypopnea syndrome; PCS, physical component summary; RCT, randomised controlled trial; SF36, Short Form 36; VAS, visual analogue scale. 3.3 HRQoL measurement tools

A total of seven different questionnaires (SF8, SF12, SF36, EQ5D, WHOQOL-BREF, WHO-5, Functional Outcomes of Sleep Questionnaire [FOSQ]) were utilised in the 30 included studies to assess HRQoL in this population. Of these questionnaires, six of these were generic, and one was a sleep-specific HRQoL questionnaire (Table 2). The most frequently used questionnaire was the SF36, which was utilised in 22 of the 30 studies (Table 1). The EQ5D was used to assess HRQoL in six studies (Dauvilliers et al., 2017, 2019; Dodel et al., 2007; Emsellem et al., 2020; Ingravallo et al., 2012; Kovalska et al., 2016), and the FOSQ was used in five studies (Dauvilliers et al., 2011; Emsellem et al., 2020; Kapella et al., 2015; Teixeira, Faccenda, & Douglas, 2004; Weaver & Cuellar, 2006). The remaining questionnaires were used in singular studies (Table 1).

TABLE 2. Meta-analysed SF36 results and associated study variables PF RP BP GH PCS PCSI V SF RE MH MCS MCSI Pooled mean 67.84 45.99 64.19 53.59 48.32 45.91 42.01 55.66 55.22 58.71 45.47 42.98 95% CI 59.26–76.42 40.80–51.17 56.87–71.50 48.12–59.05 44.45–52.20 43.01–48.81 37.22–46.79 46.56–64.75 48.79–61.66 52.14–65.27 39.97–50.97 39.02–46.95 Heterogeneity I2% 99.58 95.76 98.87 98.19 99.10 98.89 98.15 99.15 97.41 98.83 99.43 99.11 Correlators Sample size −0.217 −0.566 −0.154 0.091 −0.310 0.132 −0.434 −0.273 −0.119 1.000 ** −0.600 −0.288 Mean age 0.608

留言 (0)

沒有登入
gif