  |
|
|
||||||||||
|
||||||||||||
| Journal of the New Zealand Medical Association,
19-May-2006, Vol 119 No 1234
|
||||||||||||
|
Current and former smoking increases mortality in patients on
peritoneal dialysis
Geoffrey Braatvedt, Bronwyn Rosie, Warwick Bagg, John
Collins
Rates of cardiovascular disease in patients on renal
replacement therapy (RRT) for end-stage renal failure (ESRF) are very high.
Moreover, the survival of diabetic patients is worse than for non-diabetic
patients.1–4 Patients on RRT who develop an acute myocardial infarction
have very high 1 and 5-year mortality rates (60% and 90% respectively), which is
even higher in diabetic patients.5
Given the high rate of vascular disease in patients with
ESRF, aggressive control of modifiable risk factors such as hypertension, raised
cholesterol, and strategies to enhance smoking cessation are likely to be
especially important. We tested the hypothesis that survival in patients with
ESRF may be further adversely affected by smoking and compared survival in
patients with or without diabetes as the cause of ESRF who did or did not smoke
at peritoneal dialysis commencement.
MethodsPatients—Comprehensive
data on all patients commencing peritoneal dialysis in New Zealand has been
collected prospectively since November 1985 and entered onto a national database
(J. Collins New Zealand Peritoneal Dialysis Registry, Auckland). Cross-checks
with the Australian and New Zealand Dialysis and Transplant Registry was
undertaken on patients with missing data on the national dataset. Each
patient’s dataset was updated 6-monthly by the patient’s renal team,
and date and cause of death were also recorded from chart review.
All data are collated centrally by a data manager to
ensure accuracy of data collection - diabetes status, smoking status at
commencement of dialysis and ethnicity was known for 100% of the cohort with
only 11 patients lacking detail on vascular disease status. Smoking status at
commencement of dialysis was recorded in all centres by patient interview,
(prospective smoking status within patient groups was recorded at one centre
until 1993) and ethnicity was self-reported.
The cause of ESRF was stated at commencement of
dialysis. Patients were classified as having either type 1 or type 2 diabetes
and the diagnosis of diabetic nephropathy was made by the patient’s
nephrologists on clinical grounds. Whilst some patients with nephropathy
classified clinically as diabetic in origin may have had additional causes of
renal disease, renal biopsies were not routinely performed.
All patients who commenced peritoneal dialysis as
initial treatment for ESRF between 15 November 1985 and 15 November 1995 were
included in this analysis and followed until survey end (1 July 1997) or death.
The primary end-point of the study was death. Relationships between the
following secondary end-points and the following independent variables were
investigated: diabetes status (no diabetes, type 1 or type 2); smoking behaviour
(current, former, non); gender; ethnicity; age at peritoneal dialysis (PD)
commencement; subsequent renal replacement treatment (PD, haemodialysis,
transplant);and presence/absence of established ischaemic heart disease,
cerebrovascular or peripheral vascular disease at commencement of PD.
The complete cohort had updated data entered 6-monthly.
At commencement of dialysis, clinical history and examination (including ECG)
was used to classify patients as having “confirmed”,
“suspected”, or “no evidence” of established ischaemic
heart disease, cerebrovascular disease, or peripheral vascular disease.
Statistical
methods—Kaplan Meier survival curves and log rank tests (SAS proc
life test) were used to investigate relationships between the independent
variables and survival time. Cox proportional hazards regression (SAS proc
phreg) was used to investigate the effects of covariates. Contingency tables
(SAS proc frequency) were used to investigate relationships between the
independent variables and cause of death.
In most analyses, the outcome for patients with type 1
and type 2 diabetes were similar, despite controlling the data for other
covariates. Therefore patients with diabetes were grouped together. Similarly,
there were no significant differences between the outcomes of current and former
smokers (despite correcting for a number of covariates) so current and former
smokers were also grouped for most analyses.
It is to be noted that “former smokers”
range from those who gave up smoking many years before ESRF to those who gave up
immediately prior to dialysis. The registry did not record pack years of
smoking, nor years since stopping smoking. However, as we have defined patients
who stopped smoking many years ago as “smokers”, the group outcomes
are extremely conservative.
Classifying patients who gave up smoking more than 5 or
10 years before ESRF as “smokers”, was likely to have decreased
differences between the two groups. In the one centre that did report smoking
status at commencement of dialysis as well as prospectively, analysis showed
that smoking status at commencement of therapy for ESRF did not vary over time;
very few current smokers at commencement of RRT ceased during the follow-up, and
very few former smokers re-commenced smoking.
Inclusion in the study was based on peritoneal dialysis
as the initial mode of therapy. However the Australia and New Zealand dataset
does have outcome data recorded on all patients regardless of final mode of
therapy and thus no censoring of the data was made for patients who transferred
to other forms of treatment.
ResultsEffects of smoking on
mortality—Between November 1985 and November 1995, a total of 1293
patients commenced PD in New Zealand due to ESRF (Table 1). Follow-up data was
available on all patients for a range of 20–140 months.
Thirty-five percent of the patients had diabetic nephropathy
as the cause of ESRF. The patients with type 2 diabetes were older at
commencement of dialysis than the patients without diabetes and those with type
1 diabetes. Although current smoking prevalence at commencement of dialysis was
similar between patients with diabetes and those without diabetes (and lower
than the 1996 New Zealand non-diabetic population),6 the percentage of patients
who were lifetime non-smokers was lower in those with type 2 diabetes. As
expected, the patients with diabetes had a higher prevalence of established or
suspected macrovascular disease at onset of dialysis than those patients without
diabetes (Table 2).
After correction for age, smoking status, ethnicity, and
presence or absence of diabetes, mortality was similar in women and men (data
not shown). The survival of patients without diabetes was significantly longer
than in those with diabetes (Figure 1), even when controlling the data for age,
smoking status, and presence or absence of established cardiovascular disease at
commencement of dialysis (Table 3).
Table 3. Hazard ratio of different variables on
survival time in 1293 patients on PD for ESRF. Data is stratified by patient age
at start of PD and adjusted for gender, smoking status, ethnicity, and the
presence of macrovascular disease at treatment start
IHD=Ischaemic heart
disease; CVD=Cerebrovascular disease; PVD=Peripheral vascular
disease.
When the data was stratified for patient age at treatment
commencement (i.e. when patients of similar age at commencement of dialysis were
compared), the difference in survival between patients with diabetes and those
without diabetes was even greater (data not shown). There was no significant
difference in survival time between patients with type 1 or type 2 diabetes.
However, when patients of similar age at commencement of PD were compared, the
patients with type 1 diabetes had a shorter survival time than patients with
type 2 diabetes (data not shown).
There was a significant inverse relationship between
survival time and smoking behaviour—lifetime non smokers survived longer
than current or former smokers (Figure 2A), an effect that remained constant
after controlling for age, ethnicity, and the presence of diabetes or
macrovascular disease at commencement of PD.
Smoking and diabetes both affected survival time (Figure
2B); however those effects were additive rather than multiplicative. This
relationship remained constant even after controlling for age, ethnicity, or
macrovascular disease at commencement of PD (Table 3).
Figure 1. Kaplan Meier curves showing proportion of
patients (n=1293) commencing peritoneal dialysis between 15 November 1985 and 15
November 1995, surviving as at 1 July 1997 with type 1 diabetes (n=143), type 2
diabetes (n=308) or no diabetes (n=842). The survival of patients without
diabetes was significantly greater than those patients with diabetes
(p<0.001)
 Figure 2A. Kaplan Meier curves showing proportion of
patients (n=1293) commencing peritoneal dialysis (PD) between 15 November 1985
and 15 November 1995, surviving as at 1 July 1997 who were lifetime non smokers
(n=496), former smokers (n=578) or current smokers (n=219) at commencement of
PD. The survival of non-smokers was significantly greater than former or current
smokers (p<0.001)
 Figure 2B. Kaplan Meier
curves showing proportion of patients (n=1293) commencing peritoneal dialysis
(PD) between 15 November 1985 and 15 November 1995, surviving as at 1 July 1997,
with and without diabetes who were lifetime non smokers or current / former
smokers at commencement of peritoneal dialysis. Non-diabetic non-smokers
(n=348), non diabetic smokers (n=494), diabetic non smokers (n=145), diabetic
smokers (n=306). Each survival curve is significantly different from the others.
(p<0.001)
 Relationship between
diabetes status, smoking, and cause of death—Of the 1293 patients
in the cohort, 616 (47.6%) had died and cause of death identified by 1 July
1997. The causes of death were classified as cardiac, vascular, infective, and
“other” (includes withdrawal of dialysis). There was no significant
relationship between causes of death and smoking behaviour both before and after
correction for covariates. However, there was a strong relationship between
causes of death and diabetes, with diabetic patients more likely to die of
cardiac causes (p<0.001). Of 348 patients without diabetes who died, 46% died
of cardiac, 11% vascular, 15% infective, and 28% other causes, whereas of 268
patients with diabetes who died, 59% died of cardiac, 8% vascular, 12%
infective, and 20% other causes.
Patients with type 2 diabetes were even more likely to die
of cardiac causes than patients with type 1 diabetes and those without diabetes
(type 2 61%, type 1 56%, non-diabetic 46%). Maori and Pacific patients had
higher rates of cardiac death than Europeans (59%, 57%, 44% respectively;
p<0.005). This relationship remained after sequentially correcting the data
for diabetes, smoking, and at age PD commencement.
As expected, patients with confirmed or suspected ischaemic
heart disease (IHD) and peripheral vascular disease (PVD) at commencement of
dialysis had higher rates of subsequent cardiac death than those without those
conditions (IHD present 61% versus IHD absent 47%; PVD present 58% versus PVD
absent 50%-data not shown).
DiscussionThis study has confirmed that patients receiving PD for ESRF
due to diabetes had higher mortality than patients with ESRF not due to
diabetes.1–4 Patients with ESRF who were current (or former smokers) also
had higher mortality than non smokers. Patients with diabetes and a history of
smoking had the highest mortality of all groups.
The prevalence of smoking in patients with diabetes (and
normal renal function) compared to subjects without diabetes in New Zealand is
not well established, but is likely to be similar. In 1996, 23.7% of the adult
New Zealand population reported that they were regular smokers, with
particularly high rates in those of Maori origin (40.5%).6 Despite a 25% overall
population reduction in smoking rates from 1981 through to 1996, the prevalence
of smoking in Maori and Pacific Island people in New Zealand remained very
high.6–8
Diabetes is particularly common in Maori and Pacific Island
people, and currently is the leading single cause of ESRF in New Zealand,
accounting for about 40% of all new cases over the years 1997–2000.9
Despite Maori (14.5%) and Pacific Island people (5.6%) making up only 20.1% of
the total New Zealand population,10 76% of patients with ESRF due to diabetes
are Maori or Pacific Island people. Thus, the high rate of smoking in those
ethnic groups is particularly concerning.
Patients with diabetes who smoke may develop microvascular
complications at a higher rate including an increased risk of developing
important proteinuria.11,12 In the nurses’ health study cohort, smoking
was associated in a dose-response manner, with an increased mortality among
women with type 2 diabetes.13 The influence of smoking on outcome in patients in
ESRF (and especially in patients with diabetes) has, however, received little
study to date.
An early report on outcome of RRT in 82 diabetic patients
showed that current smoking non-significantly increased the relative risk of
death by 2.28 (0.93 to 4.84).14 Later studies of 196 diabetic patients on
haemodialysis (who were followed for 3 years),4 and 165 non-diabetic patients
compared with 118 diabetic patients on peritoneal dialysis (PD)3 surprisingly
showed no adverse effects of smoking on death rate in the diabetic patients when
compared with mortality in non-diabetic patients.
In a subsequent study of 52 patients on haemodialysis due to
diabetic nephropathy (22 who smoked and 30 who did not), survival at 1 and 5
years was significantly lower in the smoking patients (1- and 5-year survival
68% and 9% in smokers respectively; non-smokers 80% and 37% respectively).15
These studies are however either short-term or have very
small numbers of patients for follow-up. The largest report on cardiovascular
outcome in ESRF patients with and without diabetes (total number 627,983
patients) did not report the smoking status of the patients.5 A more recent
2-year study of nearly 4000 patients (44% with diabetes) on dialysis showed a
37% increase in mortality rates in patients who were current smokers compared
with non-smokers. 16
Patients in ESRF have high rates of cardiovascular
disease.17 The cause of this high risk is multifactorial and includes the
increased risks associated with the underlying cause of the ESRF (such as
hypertension and diabetes), but also the added risks caused by ESRF
itself.18–20 Patients on dialysis who suffer an acute myocardial
infarction have an especially high mortality.5
In the current study, mortality was higher in patients with
a history of smoking. Interestingly, there was no significant difference in
outcome between current smokers and former smokers, and both groups fared worse
than non-smokers. Former smokers included patients who had very recently quit
smoking prior to commencing dialysis, as well as patients with a more distant
history of regular smoking.
The fact that former smokers on dialysis had similar adverse
outcomes to current smokers suggests that smoking during advancing renal
dysfunction accelerates atherosclerosis to a degree that when added together
with the independent added burdens of ESRF on progression of atherosclerosis,
causes long-term adverse effects not corrected by stopping smoking.
Some centres in New Zealand collected prospective data on
each patient’s smoking behaviour, not only at commencement of dialysis,
but at 6-monthly intervals. These data show that very few patients who were
smoking at commencement of dialysis quit during follow-up (14.5 vs 12.8%), and
similarly that very few patients who were former smokers at commencement of
dialysis recommenced smoking during follow-up (data not shown).
The New Zealand dataset did not prospectively record
biochemical variables, and thus the independent contribution of serum lipid
concentration, fibrinogen, HbA1c (or other indices of metabolic control in
patients with diabetes) on overall outcome is unavailable. It is possible that
smokers had more unfavourable lipid profiles than non-smokers, or had more
hypertension than non-smokers. However despite earlier reports from the New
Zealand dataset showing that patients with diabetes on ESRF had higher blood
pressures than patients without diabetes, the patients with diabetes who smoked
had far worse outcomes than those patients with diabetes who did not
smoke—thus suggesting an independent adverse effect of smoking on
mortality.
It is possible that smoking behaviour is a surrogate marker
for another factor not examined in this study that does directly impact on
mortality (e.g. socioeconomic status, alcohol intake, diet, level of exercise,
lung function). The lack of difference on mortality between current and former
smokers suggests that in the context of renal dysfunction, smoking causes
irreversible effects, presumably on the endothelium with resulting permanent
adverse effects on mortality.
A larger (but only 2 year) study16 of the effects of smoking
on cardiovascular morbidity and mortality in patients on dialysis does, however,
suggest that patients with a history of having given up smoking within 1 year
before commencing dialysis had outcomes comparable to those with a lifetime
non-smoking history. This suggests that encouraging patients to quit smoking
does have long-term benefits even in the context of advanced renal disease.
In conclusion, this study has confirmed high rates of
mortality in patients on peritoneal dialysis, with diabetic patients having
higher mortality then non-diabetic patients. In addition, patients with a
history of current or former smoking had higher mortality rates then lifetime
non-smokers.
Patients with a history of diabetes and smoking had the
highest mortality of all groups.
Author information:
Geoffrey Braatvedt, Associate Professor, Endocrinologist, Department of
Medicine, University of Auckland; Bronwyn Rosie, Medical Student, University of
Auckland; Warwick Bagg, Senior Lecturer, Endocrinologist, University of
Auckland; John Collins, Renal Physician, Auckland City Hospital; Auckland
Acknowledgements:
The PD Registry Base is funded by the New Zealand Ministry of
Health. Dr W Bagg was funded by a
Diabetes New Zealand AMP Fellowship.
Correspondence:
Assoc Professor GD Braatvedt, Department of Medicine, University of Auckland,
Level 12 Auckland Hospital Support Building, Park Road Grafton, Private Bag
92-019, Auckland. Fax: (09) 367 7146; email: g.braatvedt@auckland.ac.nz
References:
|
|
||||||||||
| | Current
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads) Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals |
|