The inclusion and exclusion processes are shown in the PRISMA flowchart (Fig. 1) [11, 12]. A total of 383 unique publication abstracts, including conference abstracts, were retrieved from database searches. A review of these abstracts identified 66 articles and 9 conference posters that could have met the inclusion criteria for our review. After the full text of the 66 articles were reviewed, 52 were excluded because they did not report utility values for individuals with HD. The final number of articles and conference posters included in this systematic literature review was 16 (14 articles reporting 11 unique studies and 2 conference posters).

PRISMA diagram for study inclusion and exclusion

HRQOL = health-related quality of life, PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses

Adapted from Moher et al. [12]

Table 1 presents a list of the included studies and key features of each study. Four of the 14 articles [13,14,15,16] identified in Fig. 1 reported utility data originating from the same study. Carlozzi et al. [16] was identified as the primary report due to the richness of reported utility data, and the other 3 articles [13,14,15] were examined for any additional relevant information. In addition, two conference abstracts identified were earlier presentations of data published in subsequent articles by the same authors (Claassen et al. [17], Rodriguez Santana et al. [18]). Therefore, 11 distinct studies from the 16 identified publications were included in our analysis. Table 1 lists the characteristics of the 11 studies. Among the 11 distinct studies, 4 included participants located in the US, 2 in the UK, and 3 were multinational (2 including European countries and 1 including both US and European countries). The remaining 2 studies included participants located in Spain and Canada.

The utility measure used most frequently was the EQ-5D index (9 of 11 studies, Table 1). Two studies used the 3-level version of the instrument [19, 20], 2 studies used the 5-level version [18, 21], 1 study mapped the EQ-5D from SF-36 data [22], and the remaining 4 studies did not report the EQ-5D version. Two studies used the SF-6D to estimate utility by disease stage in patients with HD [18, 23], one of which also reported EQ-5D utility values [18]. One study did not report EQ-5D or SF-6D but instead estimated utility values associated with 4 levels of chorea severity described in vignettes using time trade-off methods in a sample of the general population [17].

A variety of disease-specific HRQOL measures were used in the studies alongside preference-based generic utility measures, including the HD-PRO-TRIAD, Quality of Life in Neurological Disorders, HDQLIFE Short Form, the Huntington Quality of Life Instrument (H-QoL-I), and the HDQoL. Also, EQ-5D visual analogue scale (VAS) scores were often reported alongside the EQ-5D index scores. For example, a cross-sectional survey study by Exuzides and colleagues [21] reported both the EQ-5D index and VAS scores.

The demographic and clinical characteristics of the populations included in the 10 studies that included patients with HD are reported in Table 1 (the 11th study, Claassen et al. [17], was a vignette valuation study in the general population). The study populations appear to be generally representative of the overall HD population, with a similar number of men and women [1]. However, we identified 3 cross-sectional surveys with more unbalanced samples (Exuzides et al. [21] with 68.3% female; Shaw et al. [22] with 28% female; Rodriguez Santana et al. [18] with 35% female). The average age of the participants was between 40 and 58 years, which appears to be consistent with other studies in HD that describe age at onset (mean = 30–50 years; range = 2–85) and disease duration (mean = 17–20 years) [24]. The verification of diagnosis differed somewhat among the studies (Table 1), with studies conducted at clinical sites using clinical diagnosis [16, 18, 23] or a positive gene status [20]. Two studies did not report details on how diagnoses were verified [19, 25], and 4 studies allowed participant self-reported diagnoses, either by genetic identification or clinical diagnosis [21, 22, 26, 27]. Only one of these studies [26] reported the proportion of patients in which the diagnosis was based on gene mutation status or clinical diagnosis: 87% of participants reported a positive gene mutation status and 67% reported clinical diagnosis. Where reported, the mean number of years since diagnosis varied from 3.97 (Carlozzi et al. [16]) to 9.0 years (Shaw et al. [22]); in 4 studies, it was not reported.

Among the 11 studies, 2 were based on site-administered surveys along with clinical observations [16, 23], and 5 were cross-sectional self-administered surveys [21, 22, 25,26,27]. Two studies were based on cross-sectional survey data collected from patients and physicians in an existing HD dataset (Rodriguez Santana et al. [18]; Huntington’s Disease Burden of Illness Study, Dorey et al. [19]; European-Huntington’s disease burden study). The remaining 2 studies included a vignette study in the general population [17] and a randomized controlled trial [20].

Among the surveys of patients with HD, more than half (7/9) were cross-sectional studies that relied on self-/proxy- or physician survey–reported outcomes completely; the remaining 2 studies [16, 23] were longitudinal, prospective, observational studies that required clinicians to collect at least some of the key clinical outcomes, such as clinical diagnoses of HD, years since HD diagnosis, staging, and severity. All 9 survey-based studies collected self-reported outcomes from patients with HD; 1 of these studies also collected proxy-rated utility values for all participants [27]. One study collected proxy-rated utility values only for patients who could not respond due to advanced disease [18].

Table 2 summarizes the utility estimates reported in each of the 11 studies identified. Among the 9 studies reporting EQ-5D utility index data, utility estimates for the overall HD population ranged from 0.81 (mean age = 48.74 years, 41.0% male, mean years since diagnosis = 3.97, prodromal HD 38.5%, early HD 38.0%, late HD 23.5%) [16] to 0.3 (mean age = 57.1 years, 45.1% male, mean years since diagnosis = 4.9) [25]. In early-stage and mid-stage disease, the anxiety and depression dimension was the main driver of poor EQ-5D utility scores, while in patients with advanced-stage disease, the main drivers were mobility, followed by the self-care and usual activities [18]. One study used the EQ-5D utility index to compare HD patients with an age- and sex-matched general population cohort [21]. In the cross-sectional survey conducted by Exuzides and colleagues in the US, the mean EQ-5D-5 L index score was significantly lower for patients with HD than for the general population (0.66 vs. 0.81; P < 0.001); almost half of the HD group had early-stage disease.

Where data were reported for different levels of severity of HD, utility scores generally worsened with increasing HD severity. In the cross-sectional US Huntington’s Disease Burden of Illness Study, Rodriguez Santana and colleagues [18] reported a decline in the mean (standard deviation [SD]) EQ-5D-5 L index score for patients with advancing disease (early-stage: 0.72 [0.22]; mid-stage: 0.62 [0.18]; and advanced-stage: 0.37 [0.30]). Carlozzi et al. [16] and Dorey et al. [19] also reported decreasing EQ-5D index scores with advancing disease.

Although all studies exploring EQ-5D values for different disease stages observed lower utility values with advancing disease, the score ranges within each disease stage varied considerably across studies. For example, in the cross-sectional study of patients with HD by Rodriguez Santana et al. [18], the mean (SD) EQ-5D index score for patients with early- (Total Functional Capacity sum of score [TFC] 13 − 7), mid- (TFC 6 − 4), or advanced-stage (TFC 3 − 0) HD, as defined by Wild et al. [28], was reported to be 0.72 (0.22), 0.62 (0.18), and 0.37 (0.30), respectively. However, Carlozzi and colleagues’ [16] longitudinal observational study reported the mean (SD) EQ-5D utility index scores for patients in prodromal, early (TFC 13 − 7), and late (TFC 6 − 0) stages of HD to be 0.89 (0.12), 0.80 (0.14), and 0.71 (0.17), respectively. The utility estimate for TFC 13 − 7 was higher in Carlozzi and colleagues (0.80 versus 0.72), and the estimate for TFC 6 − 0 in Carlozzi and colleagues was higher than the estimate for TFC 6 − 4 in Rodriguez Santana (0.71 versus 0.62), despite the latter excluding later stage patients (TFC 3 − 0) where utility was lower (0.37). Both were large studies (336 and 536 patients, respectively, with sample sizes ranging from 88 to 205 within each severity category), with clinician-verified diagnosis and similar mean age (47.3 vs. 48.74 years). There were fewer women in the Rodriguez Santana sample (35% vs. 59% in the Carlozzi sample). Some of the difference between studies may be attributed to differences in the EQ-5D version and value set used. Rodriguez Santana and colleagues [18] used the EQ-5D-5L with utility calculated using the England value set (the specific version of the England value set was not reported); Carlozzi and colleagues [16] did not report which version of the EQ-5D instrument or which value set was used. Rodriguez Santana and colleagues used proxy completion for patients with a severe cognitive deficit (the mean for all respondents is reported earlier in this paragraph); Carlozzi and colleagues [16] did not report use of proxy completion. The use of proxy completion may also explain some of the difference in utility estimates, as the proxy utility value for advanced-stage disease was much lower than the patient-reported value (0.13 versus 0.42, respectively).

A visual comparison of mean EQ-5D index utility scores across studies in patients with differing numbers of years since HD diagnosis does not reveal any obvious trend between time since HD diagnosis and mean utility value (Fig. 2). For example, in a US cross-sectional survey of patients with a mean 4.8 years since diagnosis, the mean (SD) EQ-5D index score was estimated to be 0.6 (0.3) [27]. In contrast, a Canadian cross-sectional survey [22] of patients with a substantially longer average number of years since diagnosis (9.9 years) estimated the mean (SD) EQ-5D index score to be higher (0.72 [0.24]) than that in the US survey. In both studies, HD diagnosis and time since diagnosis were self-reported. Some of the difference in utility estimates may be explained by the fact that the Canadian study [22] estimated EQ-5D utility index from SF-36 data using the mapping algorithm by Rowen et al. [29], which has been reported to overpredict utility for more severe health states. However, a regression analysis of SF-6D data by Hawton and colleagues [23] found no significant relationship between utility and time since diagnosis. The authors acknowledged some individuals are diagnosed in the premanifest stage because of family history of the condition and early predictive testing and may live for decades after diagnosis without any clinical expression, while others may be diagnosed much later at the point of significant functioning loss. Such heterogeneity in the timing of diagnosis might have made it difficult to interpret the relationship between years since diagnosis and severity levels or health statuses of patients with HD.

Comparison of published EQ-5D utility scores and mean time since diagnosis for patients with HD

HD = Huntington’s disease; EQ-5D = EuroQol 5-dimension

Two studies reported SF-6D estimates [18, 23]. In a longitudinal, observational study of patients with clinically diagnosed HD in 12 European countries, Hawton and colleagues [23] reported that SF-6D utility estimates declined with disease stage: TFC 13 − 11: 0.767 through TFC 2 − 0: 0.575. Utility estimates were not markedly lower than general population normative data for older people. For example, SF-6D utility was 0.68 compared with 0.73 for the general population in people aged 75–79 years. Rodriguez Santana et al. also reported that a decline in SF-6D utility corresponded with progression of disease stage, with early stage (TFC 13 − 7) yielding 0.61, through advanced stage (TFC 3 − 0) characterized by significant and total dependence on external care [28] experiences the most severe symptoms and require assistance in all activities of daily living, yielding 0.50. The decline in SF-6D utility with worsening TFC (0.61 to 0.50 from early to advanced stage) was smaller than the change in EQ-5D utility measured in the same study (0.72 to 0.37 from early manifest HD to advanced stage HD). A larger difference in EQ-5D utility between patients with HD and the general population was observed by Exuzides and colleagues [21] (0.66 and 0.81, respectively), suggesting that SF-6D may be less sensitive to the changes in HRQOL with HD progression than EQ-5D.

One vignette valuation study was identified [17]. In this study, health state descriptions (vignettes) were developed for different severities of HD chorea (with other manifestations remaining constant), and the health states were valued by general population participants using time trade-off methods. The mean (SD) utility values declined as chorea severity increased (mild: 0.64 [0.41]; mild/moderate: 0.48 [0.47]; moderate/severe: 0.26 [0.50]; severe: 0.07 [0.52]).

Among the 11 studies reviewed, 2 reported utility values assessed by proxy respondents on behalf of patients (Fig. 3). In a cross-sectional survey by Carlozzi et al. [27], the patient- and proxy-reported mean EQ-5D utility index scores were identical (both 0.6), although the mean proxy-reported TFC score was substantially lower than the mean patient-reported score (5.4 vs. 6.8), suggesting a disconnect in functioning capabilities by perspective. Similarly, in a cross-sectional survey by Rodriguez Santana et al. [18], proxy respondents reported similar mean EQ-5D scores as patients with mid-stage HD (0.62 [patient reported] vs. 0.66 [proxy reported]). However, large differences were reported in patients with advanced HD, where the mean (SD) score for patients was 0.42 (0.27) compared with 0.13 (0.31) for proxies. Part of this difference may be attributed to the study design in that proxy rating was used only for participants with a severe cognitive deficit.

Comparison of published utility scores for patients with HD by rater type. a. Utility Scores Measured by EQ-5D. b. Utility Scores Measured by SF-6D

HD = Huntington’s disease; EQ-5D = EuroQol 5-dimension; EQ-5D-5L = EuroQol 5-dimension 5-level; SF-6D = short-form 6-dimension; TFC = total functional capacity

Note: Error bars represent 95% confidence intervals derived from reported standard deviation and sample size (size of diamond represents the sample size). Country refers to location of residence of study participants

a Range for published mean EQ-5D and SF-6D utility for general US population of ages 35 to 64 years [49].