Liver Failure and Liver Disease
Quality of life in patients with primary biliary cirrhosis
Investigators participating in the French-PBC Study Group are listed in the appendix.
The impact of primary biliary cirrhosis (PBC) on health-related quality of life (HRQOL) is poorly documented. We assessed quality of life in a group of 276 unselected patients with PBC using the Nottingham Health Profile (NHP). This is a generic scale that assesses six major areas commonly associated with HRQOL. Data were compared with those of a sex- and age-matched control group. The associations between NHP scores and the severity of PBC were tested. Patients (86% women) had a median age of 62 years (range 33–87). Most patients were treated with UDCA. PBC patients showed a strong statistically significant difference in energy compared to controls (respectively, 40.6 vs. 22.9, P < .0001) and had worse scores for emotional reactions (22.2 vs. 16.1, P < .005). No other differences were observed. No associations of the dimension subscores were found with biochemical liver tests, histological stages, or duration of the disease. Among the signs or symptoms, fatigue was the finding most often associated with the dimension subscores. In conclusion, patients with PBC feel that their overall quality of life is worse than that of the control population. This difference is mainly due to the decrease in the subscores of energy and emotional reactions, both associated with fatigue. These effects must be taken into account by clinicians when treating these patients, as they constitute the clinical outcomes that have the most impact on patients' lifestyle and adherence to treatment. (HEPATOLOGY 2004;40:489–494.)
There has been an increasing consensus regarding the importance of patients' point of view in the assessment of health status. Traditional outcomes, important events for clinicians, need to be integrated with patients' perceptions of their health status, reflecting how they feel and how much their disease affects their way of life. This has been referred to in the literature as health-related quality of life (HRQOL). Limited information is available on the impact of primary biliary cirrhosis (PBC) on quality of life. One recent study, using approximately 100 patients with PBC or primary sclerosing cholangitis, showed substantial impairment in the HRQOL of these patients. Negative changes in HRQOL were further affected by worsening disease severity.1 Other studies have focused on specific aspects of the concept, e.g., fatigue, depression,2–4 or quality of life after liver transplantation.5, 6
Recognizing that the patient's perception of his or her quality of life is an important part of clinical care, we assessed quality of life in a large cohort of unselected patients with PBC using a generic quality-of-life scale, the French version of the Nottingham Health Profile (NHP) questionnaire.7, 8 The NHP is a generic scale that has the advantage of encompassing the six major subdimensions of quality of life and can identify those aspects affected by the disease.
The aims of this study were to (1) individualize the impact of PBC on quality of life by comparing the NHP scores of the group of patients with PBC to those of a sex- and age-matched control group; (2) look for associations between NHP scores and the main characteristics of PBC; and (3) test whether ursodeoxycholic acid (UDCA) therapy, the only medical treatment currently approved for this condition, affects quality of life in PBC.
Patients and Methods
A total of 528 patients with PBC were included in an ongoing multi-center prospective study regardless of the duration or severity of the disease. Diagnosis of PBC was based on accepted criteria. Histological staging was defined according to the Ludwig system.9 The exclusion criteria were the presence of associated disorders threatening life expectancy, previous liver transplantation, or inclusion on the waiting list for liver transplantation. Prior written, informed consent was obtained from each patient. The study received approval from the government institution allowing development of computerized files.
At entry into the study, the usual parameters that characterize the disease were recorded. Pruritus was graded as follows: absent (score of 0), occasional (1), or permanent (2). Similarly, fatigue was graded as absent (score of 0), hindering normal activity (1), or preventing normal activity (2). Symptomatic patients were defined by the presence of at least one of the following signs or symptoms: jaundice, pruritus, ascites, hepatomegaly, splenomegaly, or gastrointestinal (GI) bleeding.
In addition, quality of life was assessed among 276 patients by a self-administered questionnaire, the NHP,7 which has been validated in French.8 The NHP was chosen, as at the time of the study it was the only generic quality of life scale validated in French. At this time, the NHP2 did not exist in English or in French. There were no differences between the patients included in the quality of life study and the others in terms of age, sex ratio, age at diagnosis, age at entry into the study, and main liver biochemistries. The NHP consists of 38 dichotomous statements (yes or no), a positive answer implying a current problem in a specific area. Statements were classified into six dimensions of well-being: energy, sleep, pain, emotional reactions, social isolation, and physical mobility, and weighted to produce six scores ranging from 0 to 100 (worst). Overall quality of life was defined as the percentage of positive answers and considered to be the global score.
To our knowledge, there are still no data comparing HRQOL as estimated by the NHP with other generic scales such as the SF36 in this setting. In any case, the French version of the SF36 was not available when the study was done. It was therefore impossible for us to compare the clinical relevance of the results of the NHP with results of other generic QOL instruments.
We designed a case-control study to compare the NHP scores of patients with PBC with those of a control group matched for sex and age (within 3 years) from a nonselected population. We controlled for three parameters: age, sex, and the presence or absence of the disease. It is likely that some controls could have had co-morbid conditions such as diabetes, coronary artery disease, arthritis, etc. The picture was the same in patients with PBC. In addition to PBC, some patients presented co-morbid conditions. The frequency of these comorbid conditions was similar in patients with PBC and subjects of the same age groups (personal data). The control group was constructed as follows: (a) 173 subjects were randomly selected from a large prospective cohort, the “ESTEV cohort,” in which only workers under the age of 65 were included;10 and b) volunteers over the age of 65 (n = 103) who were asked to complete the NHP questionnaire and were either non-affected relatives of patients with PBC or relatives of the medical staff. The NHP questionnaire was given to patients and controls, in an identical format, accompanied by a short note. The main points covered were: a) the study on quality of life was part of a larger study on PBC; (b) subjects were asked to respond to the questions according to their feelings at the present time and to complete all questions; and (c) the questionnaires were anonymous.
In order to differentiate the impact of PBC or age (or their interaction) on quality of life, we classified our population into four subgroups: patients with PBC (n = 88) or controls (n = 58) less than 56 years, and patients (n = 185) and controls (n = 191) equal to or greater than 56 years of age. This cutoff corresponds to the median age at diagnosis in the group of patients with PBC.
We then looked for associations between NHP scores (global and subscores) and the main characteristics of PBC: biochemical liver tests, signs and symptoms, histological stage, duration of the disease, and duration of UDCA treatment. Because almost all of the patients were treated with UDCA, we defined as “untreated” patients the subgroup of patients that had been treated with UDCA for less than 6 months. This duration of treatment was chosen as it corresponds to the minimal treatment period required to achieve an improvement in liver laboratory tests. We then compared the NHP global and subscores among “untreated” patients to those of patients receiving long-term UDCA therapy.
The data used in this study are those for the NHP when it was first administered. Results are expressed as means (range). Clinical variables were considered as binary variables. For this analysis, pruritus was scored in two classes: absent (score of 0) versus occasional (1) or permanent (2). Similarly, fatigue was scored in two classes: absent (score of 0) or hindering normal activity (1) versus preventing normal activity (2). Comparisons were made using a Mann-Whitney or Kruskall-Wallis test. To test whether the existence of PBC or age, or both, mainly affected quality of life, a two-way ANOVA was used with a cutoff of 56 years corresponding to the median age at diagnosis. Associations between the NHP scores and the main clinical characteristics of PBC were tested using a multiple regression analysis.
Demographic and disease characteristics of the patients at entry into the study are summarized in Table 1. All degrees of severity of the disease were represented. Of the patients, 24% were symptomatic. Histological data were available for 257 patients (93%). Most of the patients were in early histological stages of the disease (66% had stages I–II) and 11% were in stage IV, i.e., cirrhosis. All but three patients were treated with UDCA.
Table 1. Characteristics of Patients With PBC at Entry Into the Study (n = 276)
|Women (%)||240 (87%)|
|Age (yr)*||62 [25–87]|
|Duration of the disease (yr)*||4.2 [0–26.9]|
|Patients treated with UDCA (n) (%)||273 (99%)|
|Duration of UDCA treatment (yr)*||3.8 [0–13.8]|
|Hepatomegaly (n) (%)||61 (23%)|
|Splenomegaly (n) (%)||25 (10%)|
|Jaundice (n) (%)||11 (4%)|
|Pruritus (n) (%)||56 (20%)|
|Fatigue (n) (%)||90 (32%)|
|Osteopenia (n) (%)||45 (16%)|
|Sjögren syndrome (n) (%)||38 (15%)|
|Ascites (n) (%)||6 (2%)|
|Variceal bleeding (n) (%)||6 (2%)|
|Hepatic encephalopathy (n) (%)||3 (1%)|
|Symptomatic patients†||64 (24%)|
|Histologic stage (n = 257)|| |
| Stage I||75 (29%)|
| Stage II||94 (37%)|
| Stage III||58 (23%)|
| Stage IV||29 (11%)|
The control group comprised 276 subjects with a median age of 59 years (24–89), and 87% were women.
Comparison of HRQOL in Patients With PBC Versus Controls.
A worse overall quality of life, expressed as the percentage of positive answers (global NHP score), was statistically different between patients with PBC and controls (22.8 vs. 17.9, respectively, P < .002). Patients with PBC showed a strong statistically significant difference in worse scores for energy and emotional reactions compared to controls (40.6 vs. 22.9, respectively, P < .0001) and (22.2 vs. 16.1, respectively, P < .005). By contrast, no significant differences were observed between patients with PBC and controls for sleep (32.0 vs. 27.0, P = .11), pain (17.0 vs. 17.2, P = .63), and physical mobility (16.3 vs. 13.1, P = .12) scores. There was a tendency towards a worse score for social isolation in patients with PBC compared to controls (14.0 vs. 10.4, P < .07).
We then assessed whether the existence of PBC or age, or both, mainly affected quality of life. Our population was classified into four subgroups: patients with PBC or controls under 56 years and patients or controls over 56 years. The presence of PBC was the main factor that worsened the energy and emotional sub scores (P < .0001 and P < .002, respectively) and to a much lesser extent social isolation (P < .03). Age over 56 years was the main factor impairing the subscores for sleep (P < .0001), pain (P < .001), and physical mobility (P < .0001).
Associations Between NHP Scores and the Characteristics of PBC.
A worse overall quality of life, expressed as the percentage of positive answers, was associated with pruritus (P < .02), jaundice (P < .01), fatigue (P < .0001), splenomegaly (P < .001), and osteopenia (P < .002). There were no associations between the subscores of each dimension and biochemical liver tests, and in particular with serum bilirubin level. There was no association between scores of each dimension and histological stages, even with stage IV (i.e., cirrhosis) or the duration of the disease. Associations between NHP scores and the clinical characteristics of PBC are summarized in Table 2. Among the subscores, energy was significantly diminished by the presence of fatigue and osteopenia. Pain was associated with the presence of hepatomegaly and osteopenia. A bad score for sleep was associated with the presence of fatigue and Sjögren syndrome. Emotional reactions were diminished by the presence of fatigue. Social isolation was also associated with fatigue. A lack of mobility was associated with the presence of fatigue, hepatomegaly, and osteopenia (Table 2).
Table 2. Associations Between the Global NHP Scores, the Domain Subscores and the Main Clinical Characteristics of Patients With PBC (n = 276)
|Global score (% positive answers)||.02||.02||<.001||—||.007||.002||—|
|Domain|| || || || || || || |
Effect of UDCA Therapy on NHP Scores in Patients With PBC.
Within the group of patients, 40 of them had never been treated or had been treated less than 6 months with UDCA, while the others had been treated for more than 6 months (median duration, 4.8 [0.5–13.8] yr). We compared the scores of each domain between patients considered as “untreated” to those receiving long-term UDCA therapy. The most marked effect was an improvement of the mean energy subscore in patients receiving long-term UDCA therapy compared to the “untreated” patients (mean score 37.7 vs. 56.0, P < .001). Long-term UDCA therapy was also associated with an improvement in the emotional reactions subscore compared to “untreated” patients (20.0 vs. 31.2, P < .02). There was no difference for the other domains.
In a large cohort of unselected patients with PBC, we showed that patients have an impaired overall quality of life compared to a sex- and age-matched control population using a generic HRQOL instrument, the NHP. The extent of impaired HRQOL differed in the various subdimensions. The more severely affected domains were energy and emotional reactions and, to a lesser extent, social isolation. This is in keeping with a previous study by Younossi et al.1 showing that patients with cholestatic diseases (PBC) and primary sclerosing cholangitis had substantial impairment of HRQOL.
We then assessed the effect of PBC or age (or their interaction) on HRQOL. The population was classified into four groups: patients with PBC or controls under 56 years and patients with PBC or controls over 56 years. Subjects older than 56 years showed impairment in scores for sleep, pain, and physical mobility. The presence of PBC was the main factor that worsened the energy and emotional subscores and to a much lesser extent the social isolation subscore. These results strengthen the proposition that PBC, itself, specifically affects the energy and emotional dimensions. The mean energy subscore in patients with PBC was 40.6 (range 0–100) compared to 22.9 (0–100) in controls. These scores are very close to those reported in both patients with cirrhosis of differing etiology (mainly viral B and/or C hepatitis or alcoholism; score around 40) and controls (score around 24) cited in an Italian study.11 This illustrates that impairment in energy is not specific to PBC but can also be found in other chronic liver diseases. Foster et al.12 assessed the HRQOL in patients with chronic viral hepatitis C (HCV) without cirrhosis using the SF36, another generic questionnaire.13 Patients with chronic HCV infection had a marked reduction in their quality of life, affecting all dimensions, but in particular, the scores related to energy. To determine whether the reduction in HRQOL was related to HCV infection or to the awareness that one had serious illness, these authors examined the SF36 scores in a small group of patients with chronic HBV infection. Patients with chronic HBV infection had a weak impairment in their HRQOL but no significant change in the energy score.12 To better understand the impact of PBC on HRQOL, we looked for associations between subscores of each domain and the main characteristics of PBC. There were no associations with biochemical liver tests, histological stages (even with stage IV, i.e., cirrhosis), or the duration of the disease. This is at variance with the findings by Younossi et al.1 who showed that impairment of HRQOL was associated with the Mayo risk scores or serum bilirubin levels. Their study population, however, was quite different from ours: a rather small sample (75 patients with PBC) and, most importantly, a very high proportion of patients with end-stage disease (61% had cirrhosis). Among signs and symptoms, fatigue was the finding most often associated with the subscores dimensions. Osteopenia was the second symptom often associated with poor scores, with a particularly high impact in the domains of energy and pain. We did not find an impact of pruritus on HRQOL, as previously reported.1 This might be explained in two ways. First, as indicated above, the severity of the disease of the patients included in the two studies was not comparable. This emphasizes the importance of population selection in studies of PBC symptomatology. Second, to control for a type I error for the multiplicity of endpoints in the present study, the analysis of each NHP subdimension was conducted using a multiple regression model. When independently testing the impact of pruritus on each domain, we found that the energy subscore was diminished by its presence (data not shown).
Three studies2–4 focused on the impact of fatigue in PBC using specific questionnaires, the fatigue impact score (FIS)14 in two and the self-rated Fatigue Assessment Instrument (FAI)15 in the other. The prevalence of fatigue varied greatly between these studies (8%–80%) and might be related to patient selection criteria. Scores of fatigue were augmented in patients with PBC but no relationships were found between these scores and any parameter of disease severity. As fatigue is often a symptom of PBC and depression, the contribution of depression to fatigue was assessed in these three studies.2–4 In contrast to previous studies,2, 3 Goldblatt et al.4 found only a weak correlation between fatigue and depression. Thus, the origin of fatigue in PBC still remains to be defined.
Our data indicate that the main effect of long-term UDCA therapy was an improvement of the mean energy subscore. While interesting, this finding should be interpreted carefully. First, the number of untreated patients is very small compared to that of treated patients even if their main characteristics were quite similar. Second, these results should be confirmed by a longitudinal, prospective study aimed at evaluating changes in subdimensions' scores. As presented below, this issue should ideally be addressed using a disease-specific questionnaire.
The NHP questionnaire allows a simple assessment of patient quality of life. The questionnaire has been widely used and is both reliable and sensitive. It has also been validated both linguistically and psychometrically in French. It is important to recognize that generic and disease-specific instruments measure different aspects of HRQOL and are complementary. This complementarity is necessary to assess the total impact of a chronic disease on HRQOL. To date, mainly generic instruments have been used to assess HRQOL of patients with chronic liver diseases. Disease-specific instruments have only recently been developed and validated in English only.16 Of course, the use of a disease-specific questionnaire would allow the longitudinal follow-up of a cohort of patients and a more specific assessment of the role of changes in treatment and treatment effects in different patient groups, overtime. The validation of a disease-specific HRQOL questionnaire, in French, should be considered.
In the present study, the choice of a generic instrument was useful as it allowed the comparison of the HRQOL of patients with PBC to that of controls and possibly, in future studies, with other chronic liver diseases. PBC has a statistically significant effect on patients' HRQOL. These effects must be taken into account by clinicians when treating these patients, as they constitute the clinical outcomes that have the most impact on patients' lifestyle and adherence to treatment.
The authors thank Dr. Francis Derrienic (Inserm Unit 170) for providing data for control subjects. They also thank the patients and volunteers who completed the NHP questionnaire.
Members of the French PBC Study Group include clinical investigators who included more than five patients and are listed according to the number of patients included: Drs. Raoul Poupon, Olivier Chazouillères, Lawrence Serfaty, and Tony Andréani, Hôpital Saint-Antoine, Paris; Stanislas Pol, Hôpital Necker-Enfants-Malades, Paris; Jean-Philippe Miguet, Hôpital Jean Minjoz, Besancon; Thong Dao, Hôpital de la Côte de Nacre, Caen; Pierre Brissot, Yves Deugnier, and Michel Messner, Hôpital de Pontchaillou, Rennes; Jean-Pierre Capron, Hôpital Nord, Amiens; Michel Doffoël, Hôpital Civil, Strasbourg; Jean-Claude Paris and Valérie Canva, Hôpital Claude Huriez, Lille; Françoise Degos and Dominique Valla, Hôpital Beaujon, Clichy; Jean-Pierre Zarski, CHU Hôpital Nord, Grenoble; Eric Lerebours and Odile Goria, Hôpital Charles Nicolle, Rouen; Catherine Buffet, CHU Kremlin-Bicêtre, Kremlin-Bicêtre; Catherine Mathieu-Chandelier, Lille; E-Alex Pariente, Pau; Bernard Filoche, Lomme; Claude Valmage, Henin Beaumont; Thierry Fontanges, Bourgoin-Jallieu; Jean Cassigneul, Toulouse; Jean-Michel Dramard, Foix; and Manuel Gameiro, Tulle. The other investigators are listed on the website of the French Association for PBC (http://www.albi-pbc.org).