Nephrol Dial Transplant (2003) 18: 126–132

Original Article

Quality of sleep and health-related quality of life in

haemodialysis patients

Downloaded from https://academic.oup.com/ndt/article-abstract/18/1/126/1809128 by Joongbu University user on 13 May 2020

Eduard A. Iliescu1,3 , Helen Coo2, Margo H. McMurray3, Carol L. Meers3, Margo M. Quinn2,
Michael A. Singer1,3 and Wilma M. Hopman2,4

1
Department of Medicine, Queen’s University, Kingston, Ontario, 2MacKenzie Health Services Research Group, Queen’s
University, Kingston, Ontario, 3End Stage Renal Disease Program, Kingston General Hospital, Kingston, Ontario and
4
Clinical Research Unit, Kingston General Hospital, Kingston, Ontario, Canada

Abstract Introduction
Background. Sleep complaints are common in haemo-
dialysis patients. In the general population, insomnia Sleep complaints are common in patients with end-
impacts negatively on health-related quality of life stage renal disease (ESRD) on dialysis and include
(HRQoL). The objective of this study was to exa- delayed sleep onset, frequent awakening, restlessness
mine the association between quality of sleep and and daytime sleepiness [1–3]. Polysomnographic stu-
HRQoL in haemodialysis patients independent of dies have documented a high prevalence of sleep
known predictors of HRQoL. disturbance in dialysis patients including obstructive
Methods. Quality of sleep was measured using the sleep apnea (OSA), periodic movement of the legs
Pittsburgh Sleep Quality Index (PSQI) and HRQoL during sleep (PMLS) and spontaneous arousals [4–6].
was measured using the Medical Outcomes Study In the general population, insomnia and OSA are
36-item Short Form (SF-36) in 89 haemodialysis associated with decreased health-related quality of
patients. life (HRQoL) [7–9]. The onset of ESRD and dialysis
Results. Sixty-three (71%) subjects were ‘poor sleepers’ impacts significantly on functional state and HRQoL
(global PSQI )5). The SF-36 mental component [10]. Factors that have been shown in various studies
summary (MCS) and physical component summary to be associated with HRQoL in dialysis patients
(PCS) correlated inversely with the global PSQI score include haemoglobin, socio-economic level, educa-
(MCS, rs 0.28, P-0.01; PCS, rs 0.45, P-0.01). tion level, dialysis schedule, race, physical exercise,
The PCS score also correlated with age (rs 0.24, comorbidity, diabetes, intermittent claudication, pre-
Ps0.02), haemoglobin (rs0.21, Ps0.048) and vious failed transplant, sex, depression and nutritional
comorbidity (rs 0.40, P-0.01), and mean PCS status [10]. Previous studies have demonstrated an
was lower in depressed subjects (26.2 vs 35.9, Ps association between sleep disturbance and physical
0.02). Subjects with global PSQI )5 had a higher and mental well-being in dialysis patients [3].
prevalence of depression, lower haemoglobin and The objectives of the present study were to deter-
lower HRQoL in all SF-36 domains. The global mine the prevalence of ‘poor sleep’ in patients with
PSQI score was a significant independent predictor ESRD on maintenance haemodialysis using a vali-
of the MCS and PCS after controlling for age, dated sleep quality questionnaire and to examine the
sex, haemoglobin, serum albumin, comorbidity and association between quality of sleep and HRQoL
depression in multivariate analysis. while controlling for known predictors of HRQoL in
Conclusions. Poor sleep is common in dialysis patients this population.
and is associated with lower HRQoL. We hypothesize
that end-stage renal disease directly influences quality
of sleep, which in turn impacts on HRQoL. Subjects and methods
Keywords: chronic renal failure; comorbidity; haemo- This was a cross-sectional study of prevalent patients under-
dialysis; quality of life; sleep going haemodialysis in the haemodialysis units associated
with Kingston General Hospital. The subjects were recruited
Correspondence and offprint requests to: Eduard Andrei Iliescu MD, from a population of haemodialysis patients already enrolled
2058 Etherington Hall, Queen’s University, Kingston, Ontario, in a 2 year longitudinal study of HRQoL. Quality of sleep
Canada, K7L 3N6. Email: eai1@post.queensu.ca was measured concurrently with the evaluation of HRQoL

# 2003 European Renal Association–European Dialysis and Transplant Association

Quality of sleep and quality of life in haemodialysis patients 127
and the other variables at the 6 month mark of the Analysis
longitudinal study. All of the variables were measured
concurrently. Patients were excluded if they were -18 years The analysis was performed using statistical software
of age, had been on dialysis for -6 months, if they were SAS1 System for Windows2 release 6.12 (SAS Institute
unable to understand English or if they were not competent Inc., Cary, NC, USA). Spearman correlation coefficients
to give informed consent. The Queen’s University Research were used to examine associations between continuous
Ethics Board approved the protocol. variables. Student’s t-test was used to compare the means
of normally distributed variables between ‘good sleepers’
(global PSQI O5) and ‘poor sleepers’ (global PSQI )5), and
Quality of sleep the Mann–Whitney U test was used for variables that were
not normally distributed. Differences among categorical
Quality of sleep was measured using the Pittsburgh Sleep variables were analysed using the x2 test or two-tailed

Downloaded from https://academic.oup.com/ndt/article-abstract/18/1/126/1809128 by Joongbu University user on 13 May 2020

Quality Index (PSQI) [11]. This self-administered question- Fisher’s exact test as appropriate. The level of significance
naire assesses quality of sleep during the previous month was as0.05 for all comparisons. Multiple linear regression
and contains 19 self-rated questions yielding seven com- with forward stepwise selection (as0.05) was performed to
ponents: subjective sleep quality, sleep latency, sleep dura- identify factors independently associated with MCS and
tion, sleep efficiency, sleep disturbances, use of sleep PCS scores. The multivariate analysis was repeated forcing
medications and daytime dysfunction. Each component is age, sex, haemoglobin, serum albumin, comorbidity and pre-
scored from 0 to 3, yielding a global PSQI score between 0 sence of depression in the model to examine the association
and 21, with higher scores indicating lower quality of sleep. between the global PSQI score and PCS and MCS scores
The PSQI is useful in identifying good and poor sleepers. A while controlling for these variables. The MCS and PCS
global PSQI score )5 indicates that a person is a ‘poor were used as the outcome variables in preference to the
sleeper’ having severe difficulties in at least two areas or individual domains of the SF-36 to limit the number of
moderate difficulties in more than three areas [11]. regressions.

Quality of life
HRQoL was measured with the Medical Outcomes Study
36-item Short Form (SF-36) [12,13]. This instrument has Results
been used extensively in populations of patients with renal
disease [10]. The SF-36 is a 36-item self-administered Univariate analysis
questionnaire that yields scores for eight domains of
HRQoL (physical functioning, role limitations physical, Of 155 patients available to enter the longitudinal
bodily pain, general health perceptions, vitality, social func- study, 32 did not meet inclusion criteria. By 6 months,
tioning, role limitations emotional and mental health) as one subject left the study, and six subjects died. The
well as two summary scores, a mental component summary remaining 116 subjects were invited to enter the
score (MCS) and a physical component summary score cross-sectional study. Twenty subjects did not com-
(PCS). Each of the eight domains is scored out of 100, with plete the PSQI and seven did not complete the SF-36.
higher scores indicating better functioning. The MCS and
PCS scores are standardized to a mean (SD) of 50 [10], with Eighty-nine subjects were included in the analysis.
scores above and below 50 indicating above and below The characteristics of the 89 subjects are shown
average functioning, respectively. in Table 1. Eighty-eight subjects were Caucasian,
one was Native Canadian and five had failed renal
allografts. The causes of renal disease were: glo-
Comorbidity merulonephritis 12, diabetic nephropathy 22, vascularu
Comorbidity was measured using the modified Charlson hypertension 25, obstruction eight, interstitial nephritis
Comorbidity Index (CCI) [14]. The CCI has been validated in six, polycystic kidney disease five and unknown 11.
dialysis patients and is a strong predictor of clinical outcomes The majority of subjects attended haemodialysis for
in this population [14]. The CCI is a composite score of 4 h three times weekly. One subject underwent slow
multiple comorbid conditions and age. Comorbid conditions nocturnal dialysis six times weekly.
are given a score ranging from 1 to 6 and a score of 1 is added The mean (SD) global and component PSQI
for each decade above 40 years of age. For the purpose scores are shown in Table 1. The global PSQI score
of this study the comorbid conditions were determined by ranged from 0 to 20, and 63 (71%) subjects were
chart review and scored accordingly; however, age was not ‘poor sleepers’ (global PSQI )5). For subjects who
included in the index in order to examine the influence of
age on HRQoL independent of comorbidity.
recorded the cause of sleep disturbance, five described
restless legs and one described trouble breathing. The
mean PSQI component scores for the study popula-
Other variables tion and for normal controls (from 11) are shown in
Figure 1.
Age, sex, cause of renal disease, time on dialysis and presence
The mean (SD) scores for SF-36 MCS, PCS and
of partner were determined by interview and chart review.
Serum albumin (Bromcresol purple method), haemoglobin HRQoL domains are shown in Table 1. The MCS
and single pool KtuV (estimated from urea reduction ratio) ranged from 19.1 to 68.7, while the PCS ranged from
were also measured. Depression was recorded as present 12.8 to 62.0. The mean PCS was statistically lower
if the subject was taking anti-depressant medication for than the standardized average of 50 (P-0.01), while
depressed mood. the MCS was not.
128 E. A. Iliescu et al.
Table 1. Characteristics of the 89 subjects included in the study Table 2. Frequencies of comorbid conditions included in the
modified Charlson Comorbidity Index among the 89 study subjects
Variable n (%) Mean (SD)
Comorbidity score Condition Number (%)

Age (years) 60.1 (16.8)

Females (n) 34 (38.2) 1 Coronary artery disease 32 (36.0)
Time on dialysis (months) 49.4 (48.1) Congestive heart failure 11 (12.4)
Living alone (n) 25 (28.1) Peripheral vascular disease 13 (14.6)
Depressed (n) 11 (12.4) Cerebrovascular disease 13 (14.6)
Haemoglobin (gul) 115.6 (11.6) Dementia 1 (1.1)
Serum albumin (gul) 37.4 (4.2) Chronic pulmonary disease 9 (10.1)
KtuV 1.79 (0.39) Connective tissue disorder 3 (3.4)

Downloaded from https://academic.oup.com/ndt/article-abstract/18/1/126/1809128 by Joongbu University user on 13 May 2020

Quality of sleep Peptic ulcer disease 40 (44.9)
Global PSQI 8.7 (4.5) Mild liver disease 2 (2.2)
Subjective sleep quality 1.21 (0.85) Diabetes 26 (29.2)
Sleep latency 1.43 (1.16) 2 Hemiplegia 1 (1.1)
Sleep duration 1.10 (1.07) Severe renal disease 89 (100)
Sleep efficiency 1.35 (1.28) Diabetes with end-organ damage 26 (29.2)
Sleep disturbance 1.45 (0.62) Any tumour, leukaemia, lymphoma 17 (19.1)
Use of sleep medications 1.09 (1.35) 3 Moderate or severe liver disease 2 (2.2)
Daytime dysfunction 1.08 (0.69) 6 Metastatic solid tumour 0 (0.0)
Quality of life AIDS 0 (0.0)
MCS 48.8 (11.5)
PCS 34.7 (12.6)
Physical functioning 49.6 (31.5)
Role-physical 36.2 (42.6) Table 3. Correlation coefficients for the SF-36 MCS and PCS and
Bodily pain 60.2 (31.3) the other continuous variables among the 89 study subjects
General health 46.7 (25.1)
Vitality 43.1 (23.5) Variable MCS PCS
Social functioning 63.5 (44.6)
Role emotional 63.7 (44.6) r P r P
Mental health 73.4 (20.1)

MCS, SF-36 Mental Component Summary. PCS, SF-36 Physical Age 0.01 0.96 0.24 0.02
Component Summary. PSQI, Pittsburgh Sleep Quality Index. Time on dialysis 0.21 0.052 0.13 0.24
Haemoglobin 0.12 0.27 0.21 0.048
The mean (SD) CCI was 4.45 (2.1) with a range of Serum albumin 0.01 0.95 0.18 0.08
KtuV 0.03 0.75 0.10 0.33
2–11. The frequencies of comorbid conditions included CCI 0.08 0.46 0.40 -0.01
in the CCI are shown in Table 2. Global PSQI 0.28 -0.01 0.45 -0.01
The age and sex distribution of the 27 subjects Subjective sleep quality 0.21 0.046 0.20 0.06
who did not complete the PSQI or SF-36 question- Sleep latency 0.06 0.56 0.14 0.19
Sleep duration 0.00 0.98 0.08 0.43
naires were similar to the study population. The mean Sleep efficiency 0.10 0.36 0.38 -0.01
(SD) age was 61.4 (14.2) years, and eight (29.6%) were Sleep disturbance 0.27 -0.01 0.48 -0.01
female. Use of sleep medications 0.32 -0.01 0.39 -0.01
Daytime dysfunction 0.51 -0.01 0.36 -0.01

Bivariate analysis r, correlation coefficient. P, P-value for the correlation. MCS, SF-36
Mental Component Summary. PCS, SF-36 Physical Component
The correlations between MCS and PCS and the Summary. CCI, modified Charlson Comorbidity Index. PSQI,
other continuous variables are shown in Table 3. Pittsburgh Sleep Quality Index.

Fig. 1. Mean scores for components of the PSQI for the 89 study subjects and for normal controls. Data for normal controls is from Buysse
et al. [11].
Quality of sleep and quality of life in haemodialysis patients 129
Table 4. Correlation coefficients for the PSQI and the other Table 5. Characteristics of good sleepers compared to poor sleepers
continuous variables in the 89 study subjects among the 89 study subjects

Variable Global PSQI Variable Good sleepers Poor sleepers P

Global PSQI Global PSQI
O5 ns26 )5 ns63
r P

Age (years) 59.4 (19.1) 60.34 (16.0) 0.98

Age 0.05 0.61
Females (n) 6 28 0.09
Time on dialysis 0.16 0.15
Time on dialysis 50.0 (61.7) 49.1 (41.8) 0.36
Haemoglobin 0.27 -0.01
(months)
Serum albumin 0.24 0.02
Living alone (n) 6 19 0.61

Downloaded from https://academic.oup.com/ndt/article-abstract/18/1/126/1809128 by Joongbu University user on 13 May 2020

KtuV 0.11 0.30
Depressed (n) 0 11 0.03
CCI 0.03 0.80
Haemoglobin (gul) 119.2 (11.1) 114.1 (11.6) 0.03
MCS 0.28 -0.01
Serum albumin (gul) 37.9 (4.0) 37.2 (4.3) 0.27
PCS 0.45 -0.01
KtuV 1.71 (0.17) 1.82 (0.45) 0.11
Physical functioning 0.37 -0.01
CCI 4.35 (2.31) 4.49 (2.02) 0.57
Role-physical 0.45 -0.01
MCS 53.2 (8.9) 47.0 (12.0) 0.04
Bodily pain 0.54 -0.01
PCS 41.5 (11.2) 31.8 (12.1) -0.01
General health 0.32 -0.01
Physical functioning 66.9 (29.6) 42.5 (29.7) -0.01
Vitality 0.36 -0.01
Role-physical 57.7 (44.6) 27.4 (38.8) -0.01
Social functioning 0.34 -0.01
Bodily pain 78.6 (22.3) 52.6 (31.4) -0.01
Role-emotional 0.40 -0.01
General health 56.1 (27.0) 42.8 (23.3) 0.046
Mental health 0.31 -0.01
Vitality 52.5 (26.9) 39.2 (21.0) 0.02
Social functioning 75.5 (28.4) 58.5 (31.3) 0.02
r, correlation coefficient. P, P-value. MCS, SF-36 Mental Role-emotional 88.5 (28.2) 53.4 (46.2) -0.01
Component Summary. PCS, SF-36 Physical Component Summary. Mental health 80.8 (18.3) 70.3 (20.1) 0.01
CCI, modified Charlson Comorbidity Index. PSQI, Pittsburgh Sleep
Quality Index.
Results are mean (SD) unless otherwise specified. MCS, SF-36
Mental Component Summary. PCS, SF-36 Physical Component
There was significant inverse correlation between MCS Summary. CCI, modified Charlson Comorbidity Index. PSQI,
Pittsburgh Sleep Quality Index.
and global PSQI score. There was significant correla-
tion between PCS and haemoglobin, and inverse
correlations between PCS and age, CCI, and global PSQI score (bs 1.24, P-0.01). The global PSQI
PSQI score. For categorical variables, the mean PCS score remained a significant independent predictor
was lower in those with depression compared with of MCS and PCS after controlling for age, sex,
those without (PCS 26.2 vs 35.9, Ps0.02). The mean haemoglobin, serum albumin, CCI and presence of
MCS and PCS were not different for females vs depression as shown in Table 6.
males, or for subjects who lived alone vs those with
a partner.
The correlations between the global PSQI score Discussion
and the other continuous variables are shown in
Table 4. There were significant inverse correlations
of global PSQI score with haemoglobin, serum albu- The prevalence of poor sleep in the present study
min, MCS and PCS. For categorical variables, the was 71%, comparable with the 50–80% prevalence of
mean global PSQI score was higher in those with sleep-wake complaints in dialysis patients reported in
depression compared with those without (12.36 vs 8.19, previous studies [1–3]. There was a strong association
P-0.01). The mean global PSQI score was not dif- between quality of sleep and mental and physical
ferent for females vs males, or for subjects who lived HRQoL that persisted after controlling for known
alone vs those with a partner. predictors of HRQoL. Williams et al. [3] examined
The characteristics of ‘good sleepers’ (global PSQI the associations between seven specific sleep distur-
O5) compared with ‘poor sleepers’ (global PSQI )5) bances and a large number of mental, physical, func-
are shown in Table 5. Compared with ‘good sleepers’, tional and laboratory variables in 242 haemodialysis
‘poor sleepers’ had a greater proportion of depressed patients and found physical and mental well-being
subjects, lower haemoglobin and lower HRQoL in were related to the sleep disturbances. For example,
all domains. The mean SF-36 domain scores for functional status measured by the performance of
‘good sleepers’, ‘poor sleepers’ and ageusex matched activities of daily living was associated with waking
Canadian norms (from 15) are shown in Figure 2. up during the night, feeling tired in the morning and
restless sleep, while perception and memory were
associated with waking up too early. In the present
study, mental HRQoL was associated with subjective
Multivariate analysis
sleep quality, sleep disturbance, use of sleep medica-
The only significant predictor of MCS was the global tions and daytime dysfunction, while physical HRQoL
PSQI score (bs 0.852, P-0.01). The significant was associated with sleep efficiency, sleep disturbance,
independent predictors of PCS were age (bs 0.136, use of sleep medications and daytime dysfunction.
Ps0.04), CCI (bs 2.02, P-0.01) and the global Compared with ‘good sleepers’, ‘poor sleepers’ had
130 E. A. Iliescu et al.

Downloaded from https://academic.oup.com/ndt/article-abstract/18/1/126/1809128 by Joongbu University user on 13 May 2020

Fig. 2. Mean scores for the domains of the SF-36 for poor sleepers (global PSQI )5), good sleepers (global PSQI O5), and ageusex matched
Canadians from Hopman et al. [15]. MCS, SF-36 Mental Component Summary. PCS, SF-36 Physical Component Summary. PSQI,
Pittsburgh Sleep Quality Index.

Table 6. Multiple linear regression models with outcome variables positive airway pressure (nCPAP) [8,9]. We hypothe-
SF-36 MCS and PCS size that specific ESRD-related sleep disturbances
such as OSA and PMLS impact directly on HRQoL
Variable Model 1 Outcome Model 2 Outcome
MCS Adj. R2s0.08 PCS Adj. R2s0.33
and are responsible for a component of the associa-
tion between sleep quality and HRQoL observed in
the present study.
b P b P
In the present study, HRQoL was associated with
age, haemoglobin, comorbidity and depression in the
Intercept 56.9 -0.01 54.3 -0.01
Age 0.0228 0.76 0.153 0.03
bivariate analysis. These findings are consistent with
Sex 0.0528 0.98 2.91 0.23 previous studies [10]. Quality of sleep was associated
Haemoglobin 0.124 0.30 0.0331 0.76 with haemoglobin, serum albumin and depression in
Serum albumin 0.266 0.43 0.112 0.72 the bivariate analysis. Correction of anaemia with
CCI 0.804 0.21 2.01 -0.01 erythropoietin reduces PMLS, arousals from sleep and
Depression 1.76 0.66 0.935 0.80
PSQI 0.885 -0.01 1.21 -0.01 sleep fragmentation [17]. Previous studies have not
found an association between serum albumin and
b, regression coefficient. P, P-value. MCS, SF-36 Mental Component quality of sleep; however, major depression has been
Summary. PCS, SF-36 Physical Component Summary. CCI, associated with hypoalbuminaemia as part of an
modified Charlson Comorbidity Index. PSQI, Pittsburgh Sleep
Quality Index.
acute-phase response [18]. In the present study, quality
of sleep remained a significant predictor of mental
and physical HRQoL after controlling for age, sex,
lower HRQoL in all domains. The association between haemoglobin, serum albumin, comorbidity and depres-
sleep quality and HRQoL may be explained by a sion, suggesting that the relationship between quality
direct influence of sleep quality on HRQoL (or vice of sleep and HRQoL was independent of these
versa), an association of both constructs with one or potential confounding variables.
more confounding variables, an overlap in the instru- Dialysis adequacy measured by small solute clear-
ments used to measure sleep quality and HRQoL, or a ance, KtuV urea, was not associated with quality of
combination of these. sleep or HRQoL. These findings are consistent with
The best evidence that ESRD can directly influ- previous studies. Holley et al. [1] measured sleep distur-
ence quality of sleep, which in turn leads to reduced bance in 48 haemodialysis patients and 22 peritoneal
HRQoL, comes from studies of OSA in dialysis dialysis patients and found KtuV did not predict
patients. OSA is common in dialysis patients [4,5]. reported sleep disturbances. Williams et al. [3] found
Slow nocturnal dialysis and transplantation improve no association between the seven specific sleep dis-
or reverse OSA [4,16]. In patients without renal turbances and KtuV. Morton et al. [19] measured
disease, OSA is associated with reduced HRQoL HRQoL using the RAND 36 Item Health Survey
measured by the SF-36, and HRQoL improves in 115 dialysis patients and found no association
dramatically with treatment with nasal continuous between HRQoL and KtuV. The authors hypothesized
Quality of sleep and quality of life in haemodialysis patients 131

that HRQoL is influenced by factors other than the evaluation a large number of conditions while
dialysis adequacy. It may not be possible to detect limiting comorbidity to only one independent variable.
the influence of KtuV on sleep quality and HRQoL This does not necessarily mean that the index includes
within the narrow range of KtuV achieved with thrice all conditions that are important in terms of HRQoL
weekly dialysis. Alternatively, KtuV may not be a or that the value assigned to each condition based on
suitable measure of dialysis adequacy in regards to clinical outcome data is appropriate for the outcome
quality of sleep or HRQoL, and frequency of dialysis of HRQoL. Depression was recorded as present if
may be more important than quantity of dialysis. the subject was taking anti-depressant medication for
Dialysis adequacy may have a significant influence depressed mood. This definition would have misclassi-
on quality of sleep when thrice weekly dialysis is fied subjects with new onset of depression and treated
compared with daily or nocturnal dialysis. Hanly subjects who were no longer depressed. Despite the

Downloaded from https://academic.oup.com/ndt/article-abstract/18/1/126/1809128 by Joongbu University user on 13 May 2020

and Pierratos performed polysomnography in 14 potential limitations in the evaluation of comorbidity
haemodialysis patients on conventional intermittent and depression, both variables were associated with
haemodialysis before and after conversion to noctur- HRQoL in the bivariate analysis.
nal haemodialysis. Before conversion to nocturnal Further to the discussion of potential confounding
dialysis, there was a high prevalence of OSA and variables in the association between quality of sleep
PMLS [4]. In the seven subjects with OSA, conversion and HRQoL, the reviewers commented that a number
to nocturnal haemodialysis was associated with a of other variables should be considered including
dramatic reduction in respiratory arousals from 25 to lifestyle habits, income, education, diabetes indepen-
6uh accompanied by a significant rise in serum bicar- dent of the CCI, nutrition, anti-hypertensive medica-
bonate. The authors hypothesized that OSA in tions, functional status, psychiatric syndromes and
dialysis patients is due to central destabilization of daily life stress. To examine the influence of these
ventilatory control and upper airway obstruction variables on the results of this study, the analysis was
related to acidosis and airway oedema, respectively, repeated including variables for which prospective
both of which are improved with nocturnal dialysis. data was available. Forty-one subjects reported the
In the present study, quality of sleep and HRQoL perception that their income was sufficient, 34 had
were measured using validated questionnaires. The post-secondary education, 26 were diabetic, 20 were
PSQI and SF-36 evaluate quality of sleep and HRQoL current smokers, 57 used anti-hypertensive medica-
during the preceding 4-week period. The SF-36 has tions. There were no statistical differences in the
been rigorously evaluated as a tool for the measure- mean MCS, PCS and PSQI among the categories of
ment of HRQoL in patients with OSA [8,9], and has these categorical variables. This may have been in part
been used to measure HRQoL in patients with due to limited statistical power as there was a trend
insomnia [7]. The SF-36 and PSQI do not ask the to higher MCS in subjects with post-secondary educa-
same questions, however, some overlap is likely, tion compared with those without (51.4 vs 46.7,
particularly with regard to daytime dysfunction Ps0.07). Forcing each of these variables into the
such as feeling tired. This would result in an over- final models in Table 6, individually or as a group,
estimation of the relationship between this com- did not significantly change the regression coefficients
ponent of the PSQI and HRQoL. In contrast, PSQI or P-values for the PSQI. These results do not exclude
components such as sleep efficiency and sleep distur- the possibility that the relationship between quality of
bance are evaluated by specific questions regarding sleep and HRQoL is confounded by variables that
sleep times and disturbances. The strong associations were not measured (exercise, caffeine, alcohol, func-
between these PSQI components and physical HRQoL tional status, psychiatric syndromes, daily life stress),
suggest that the relationship between sleep quality or by variables that were measured with limitations
and HRQoL is not simply due to potential overlap including comorbidity and depression.
between the two questionnaires in regard to daytime The prevalent study population with little ethnic
dysfunction. diversity limits the generalizability of the results of
The main limitation of this study is the absence of this study to other populations; however, the fact that
polysomonographic data without which it is not possi- the prevalence of ‘poor sleep’ in the present study is
ble to ascertain the exact causes of insomnia and sleep similar to the prevalence of sleep-wake complaints in
disturbance. Because of the cross-sectional design it previous studies suggests that the magnitude of sleep
is not possible to establish cause and effect in the problems in haemodialysis patients is similar among
associations examined. The study aimed to control different populations.
for known potential confounding variables of the In conclusion, the results of this study suggest that
relationship between quality of sleep and HRQoL. It is ‘poor sleep’ is common in dialysis patients and that
not possible to accurately measure all variables that quality of sleep is an independent predictor of
may impact on quality of sleep and HRQoL, and the HRQoL. We hypothesize that a component of this
sample size of the study would limit the examination association is due to a direct influence of ESRD on
of numerous independent predictors of HRQoL. quality of sleep, which in turn influences HRQoL.
Comorbidity was measured with a validated index OSA serves as feasible model for this hypothesis. In
that is a strong predictor of clinical outcomes in dialysis patients with low HRQoL, measurement of
dialysis patients. The advantage of using an index is quality of sleep in conjunction with specific questions
132 E. A. Iliescu et al.

about symptoms of OSA and PMLS may be useful in 8. Jenkinson C, Stradling J, Petersen S. Comparison of three
identifying patients who would benefit from formal measures of quality of life outcome in the evaluation of
continuous positive airways pressure therapy for sleep apnoea.
polysomnography. Additional studies are needed to J Sleep Res 1997; 6: 199–204
examine the influence of objective sleep disturbances 9. Jenkinson C, Stradling J, Petersen S. How should we evaluate
on HRQoL in dialysis patients and to evaluate health status? A comparison of three methods in patients
potential treatments such as more frequent dialysis, presenting with obstructive sleep apnoea. Qual Life Res 1998;
sleep medications, anaemia management and nCPAP. 7: 95–100
Longitudinal studies of quality of sleep in patients 10. Valderrabano F, Jofre R, Lopez-Gomez JM. Quality of life in
end-stage renal disease patients. Am J Kidney Dis 2001;
with progressive renal insufficiency not yet on dialy- 38: 443–464
sis are needed to determine at what stage of renal 11. Buysse DJ, Reynolds CF 3rd, Monk TH, Berman SR, Kupfer
insufficiency quality of sleep declines and to examine

Downloaded from https://academic.oup.com/ndt/article-abstract/18/1/126/1809128 by Joongbu University user on 13 May 2020

DJ. The Pittsburgh Sleep Quality Index: a new instrument
the temporal sequence with the decrease in HRQoL. for psychiatric practice and research. Psychiatry Res 1989;
28: 193–213
12. Ware JE Jr, Snow KK, Kosinski M, Gandek B. SF-36 Health
Survey manual and interpretation guide. The Health Institute,
Acknowledgements. This study received financial support through
New England Medical Center, Boston, MA, 1993
a grant from The Kidney Foundation of Canada.
13. Ware JE Jr, Kosinski M, Keller SD. SF-36 physical and mental
summary scales: a user’s manual. The Health Institute, New
England Medical Center, Boston, MA, 1994
14. Beddhu S, Bruns FJ, Saul M, Seddon P, Zeidel ML. A simple
comorbidity scale predicts clinical outcomes and costs in dialysis
References patients. Am J Med 2000; 108: 609–613
15. Hopman WM, Towheed T, Anastassiades T et al. Canadian
1. Holley JL, Nespor S, Rault R. A comparison of reported sleep normative data for the SF-36 health survey. Canadian
disorders in patients on chronic hemodialysis and continuous Multicentre Osteoporosis Study Research Group. Can Med
peritoneal dialysis. Am J Kidney Dis 1992; 19: 156–161 Assoc J 2000; 163: 265–271
2. Walker S, Fine A, Kryger MH. Sleep complaints are common in 16. Auckley DH, Schmidt-Nowara W, Brown LK. Reversal of sleep
a dialysis unit. Am J Kidney Dis 1995; 26: 751–756 apnea hypopnea syndrome in end-stage renal disease after
3. Williams SW, Tell GS, Zheng B, Shumaker S, Rocco MV, Sevick kidney transplantation. Am J Kidney Dis 1999; 34: 739–744
MA. Correlates of sleep behavior among hemodialysis patients.
17. Benz RL, Pressman MR, Hovick ET, Peterson DD. A
Am J Nephrol 2002; 22: 18–28
preliminary study of the effects of correction of anemia with
4. Hanly PJ, Pierratos A. Improvement of sleep apnea in patients
with chronic renal failure who undergo nocturnal hemodialysis. recombinant human erythropoietin therapy on sleep, sleep
N Engl J Med 2001; 344: 102–107 disorders, and daytime sleepiness in hemodialysis patients (The
5. Kimmel PL, Miller G, Mendelson WB. Sleep apnea syndrome in SLEEPO study). Am J Kidney Dis 1999; 34: 1089–1095
chronic renal disease. Am J Med 1989; 86: 308–314 18. Maes M, Wauters A, Neels H et al. Total serum protein and
6. Benz RL, Pressman MR, Hovick ET, Peterson DD. Potential serum protein fractions in depression: relationships to depressive
novel predictors of mortality in end-stage renal disease patients symptoms and glucocorticoid activity. J Affect Disord 1995;
with sleep disorders. Am J Kidney Dis 2000; 35: 1052–1060 34: 61–69
7. Leger D, Scheuermaier K, Philip P, Paillard M, Guilleminault C. 19. Morton AR, Meers C, Singer MA et al. Quantity of
SF-36: evaluation of quality of life in severe and mild insomniacs dialysis: quality of life–what is the relationship? ASAIO J
compared with good sleepers. Psychosom Med 2001; 63: 49–55 1996; 42: M713–M717

Received for publication: 15.5.02

Accepted in revised form: 30.8.02