Anda di halaman 1dari 10

| |

Received: 30 October 2017    Revised: 8 February 2018    Accepted: 13 February 2018

DOI: 10.1111/jch.13255


Proteinuria and risk of stroke in patients with hypertension:

The Kailuan cohort study

Anxin Wang PhD1,2,3,4,5,6 | Liye Dai MD1 | Zhaoping Su MD1,2,3,4 | Shuohua Chen MD7 | 

Junjuan Li MD8 | Shouling Wu MD7 | Yongjun Wang MD1,2,3,4  | 
Yilong Wang MD, PhD1,2,3,4

Department of Neurology, Beijing Tiantan
Hospital, Capital Medical University, Beijing, Proteinuria is associated with stroke, but the effects of changes in proteinuria on
China stroke risk are not well understood in the hypertensive population. This study exam-
China National Clinical Research Center for ined whether proteinuria changes across 2-­year assessments were associated with
Neurological Diseases, Beijing, China
3 incident stroke in individuals with hypertension. We used visit data from 24 300 par-
Center of Stroke, Beijing Institute for Brain
Disorders, Beijing, China ticipants with hypertension of the Kailuan study who were stroke free at baseline.
Beijing Key Laboratory of Translational Based on the baseline and 2-­year dipstick screening results, participants were classi-
Medicine for Cerebrovascular Disease,
Beijing, China
fied as having no, remittent, incident, or persistent proteinuria. The relationship be-
Beijing Municipal Key Laboratory of Clinical tween proteinuria and stroke was analyzed using Cox proportional-­hazards models
Epidemiology, Beijing, China after adjusting for potential variables. During a median of 6.89-­year follow-­up, we
Department of Epidemiology and Health
identified 1197 people with stroke. Compared to those with no proteinuria, stroke
Statistics, School of Public Health, Capital
Medical University, Beijing, China risk was significantly increased in participants with incident (hazard ratio [HR] 1.41,
Department of Cardiology, Kailuan 95% CI, 1.05-­1.77) and persistent proteinuria (HR 1.49, 95% CI, 1.25-­1.89) after ad-
Hospital, North China University of Science
and Technology, Tangshan, China
justment for other factors, which was consistent in ischemic stroke and intracerebral
Department of Nephrology, Kailuan hemorrhage. No interaction was found between changes of proteinuria and diabetes
Hospital, North China University of Science mellitus in the hypertensive population. Changes in proteinuria exposure, particularly
and Technology, Tangshan, China
persistent proteinuria, play a role in reflecting the risk of stroke in patients with
Correspondence hypertension.
Yilong Wang, MD, PhD, Department of
Neurology, Beijing Tiantan Hospital, Capital
Medical University, Beijing, China.

1 |  I NTRO D U C TI O N group, 33.9%; normotensive, 10%).7 Furthermore, data from the China
National Stroke Screening Survey (CNSSS) showed that hypertension
Studies demonstrated that a decreased level of kidney function is an in- ranked in first place among factors contributing to stroke in China.8
dependent risk factor for all-­cause mortality as well as adverse cardio- The association between proteinuria and stroke, in particular, should
vascular disease outcomes including myocardial infarction, stroke, and be noted in the hypertensive population who are at high risk for the
progression of heart failure.1-4 As an important marker of chronic kid- development of stroke. More important, proteinuria often changes dy-
ney disease (CKD), proteinuria was proved to be associated with inci- namically, showing regression or progression.9 However, an inherent
dent stroke during the last few decades.5 Data suggested that baseline limitation of previous studies is the reliance on a single time point by
proteinuria confers a 50%-­92% greater risk of future stroke. Previous 6 which to assess kidney damage and events using baseline proteinuria.
study demonstrated that the prevalence of proteinuria was higher in The value of multiple dipstick measurements for changes in proteinuria
the hypertensive group than in the normotensive group (hypertensive and its relation with stroke outcomes has rarely been investigated.

Anxin Wang and Liye Dai contributed equally to the manuscript.

J Clin Hypertens. 2018;1–10. ©2018 Wiley Periodicals, Inc. |  1  
2       WANG et al.

Therefore, we conducted this study to assess the relationship proteinuria on the basis of easy accessibility and low cost.14 Dipstick
between proteinuria change and stroke incidence in participants positive proteinuria, seen as a sign of chronic kidney disease (CKD),
with hypertension. has emerged as a powerful and independent predictor of cardiovas-
cular disease.6 In this study, proteinuria was screened by dipstick
strictly following the direction of the kit. A freshly obtained urine
2 |  M E TH O DS
sample was first visually examined for 1 minute by a trained physi-
cian (H12-­MA, DIRUI N-­600).15 Using a color scale, the results of
2.1 | Study population
the urine strip test were semiquantified as absent, trace, 1+, 2+, or
The design, method, rationale, and examination details of the Kailuan 3+. Proteinuria positive was defined as trace or more with dipstick
cohort study have been published previously. In brief, the Kailuan analysis either at baseline or at the follow-­up visit.
study is a population-­based prospective cohort study. Between June We defined 4 proteinuria groups according to changes in protein-
2006 and October 2007, a total of 101 510 individuals aged 18-­98 uria from baseline to the follow-­up visit. Based on the baseline and
were recruited to participate in the study and were followed up until 2-­year dipstick screening results, participants were classified as having
present by means of biennial face-­to-­face examinations and annual no, remittent, incident, or persistent proteinuria. “No proteinuria” was
comprehensive surveillance of medical records and death certificates. defined as proteinuria absent both at the baseline visit and at the fol-
All participants recruited underwent routine history and physical ex- low-­up visit. “Remittent proteinuria” was defined as proteinuria present
amination, anthropometry, and laboratory assessment. at the baseline but absent at the follow-­up visit. “Incident proteinuria”
Out of the whole cohort at the first examination, 42 698 hy- was defined as proteinuria absent at the baseline visit but occurring at
pertensive participants without previous stroke and with complete the follow-­up visit. “Persistent proteinuria” was defined as proteinuria
dipstick assessment of proteinuria participated in the second exam- present at the baseline visit and also at the follow-­up visit. Reduction in
ination of the Kailuan study in the year 2008-­2009. Hypertension proteinuria was defined as baseline minus 2-­year proteinuria.
was defined as systolic blood pressure (SBP) at least 140 mm Hg or
diastolic blood pressure (DBP) at least 90 mm Hg or history of hy-
2.3 | Potential covariates
pertension and/or use of antihypertensive agents. After exclusion
for new finding of stroke during the period 2006-­2007 to 2008-­ The variables involved in this research including age, gender, current
2009 or missing proteinuria data, 24 300 participants remained smoker, current alcohol, physical activity, body mass index, diabetes
eligible for the present analysis. A flowchart of the present cohort mellitus, dyslipidemia, total cholesterol, triglycerides, low-­density li-
study is shown in Figure 1. All participants provided informed con- poprotein, high-­density lipoprotein, systolic blood pressure, diastolic
sent. The study was performed according to the guidelines of the blood pressure, fasting blood glucose, and estimated glomerular fil-
Helsinki Declaration and was approved by the Ethics Committee of tration rate (eGFR). All these variables were mentioned and described
the Kailuan Hospital and the Beijing Tiantan Hospital. in previous studies. Data on smoking, drinking, physical activity, and
other demographic variables including sex and age were collected
using questionnaires. SBP and DBP were measured twice in the seated
2.2 | Measurements of proteinuria
position using a mercury sphygmomanometer. The average of the 2
At the baseline examination, a dipstick urine analysis was performed. readings was used for the analyses. The change of blood pressure
The urine dipstick is a widely used tool in clinical screening for was calculated by diastolic in 2008-diastolic in 2006. We included this

F I G U R E   1   Flowchart of the study

WANG et al. |

variable to decrease the influence of progression by blood pressure. ratio and 95% confidence interval (CI) for stroke outcomes by changes in
Weight and height were measured during the interview and body mass proteinuria, using participants with no proteinuria as the reference group.
index (BMI) was calculated as weight (kg)/height (m) ². Blood pressure Variables with a P value < .2 and the well-­established predictors were se-
was measured with a standard mercury sphygmomanometer by a phy- lected as adjustment covariates into the multivariable analyses.
sician or trained nurses. Blood samples were collected from the an- We fitted 3 multivariable Cox proportional hazards models.
tecubital vein after an overnight fast. All blood samples were tested Model 1 was adjusted for age and sex. Model 2 was further adjusted
using a Hitachi 747 auto-­analyzer (Hitachi, Tokyo, Japan) at the central for current smoker, current alcohol user, physical activity, and body
laboratory of the Kailuan General Hospital for serum total cholesterol, mass index. Model 3 was adjusted for variables in model 2 plus di-
low-­density lipoprotein cholesterol (LDL-­C), high-­density lipoprotein abetes mellitus, dyslipidemia, total cholesterol, high-­density lipo-
cholesterol (HDL-­C), fasting blood glucose (FBG), and other factors. protein, systolic blood pressure, change of blood pressure, fasting
Previous history of disease, including hypertension, diabetes mellitus, blood glucose, and eGFR. We used these models with a sandwich
and hyperlipidemia, was collected by self-­report. The use of medicine covariance matrix as a random effect to account for the potential
such as antihypertensives within the past 2 weeks before the baseline confounding effect of the 11 different hospitals in the study. In the
interview was also self-­reported. analysis of stroke risk for per-­degree decrease of proteinuria, base-
line proteinuria was adjusted in addition to the three models.
To test the robustness of our findings, we conducted model 3 in
2.4 | Follow-­up and stroke assessment
participants excluding individuals with eGFR < 30 mL/min/1.73 m² in
We considered the 2008 survey as the starting point and December 2008 as sensitive analysis. Furthermore, we tested interactions between
31, 2015 as the endpoint of the follow-­up in the present prospec- changes in proteinuria with diabetes mellitus and sex for the risk of stroke.
tive study. Follow-­up evaluations included biennial measurements of
laboratory parameters and the occurrence of adverse events.
The outcome was the first occurrence of stroke, either the first 3 | R E S U LT S
nonfatal stroke event or stroke death without a preceding nonfatal
event. A nonfatal stroke is defined as the sudden onset of a focal A total of 24 300 participants with available data of proteinuria
neurological deficit with a vascular mechanism lasting > 24 hours, (men, 82.35%) were analyzed in our study. The mean age of the re-
including hemorrhagic stroke, ischemic stroke, and subarachnoid maining population was 55.04 ± 10.88 years. We divided the par-
hemorrhage. Stroke is diagnosed according to the World Health ticipants into 4 categories according to changes in proteinuria from
Organization criteria combined with a brain computed tomography 2006 to 2008. The baseline characteristics in both 2006 and 2008
(CT) or magnetic resonance (MR) for confirmation16 and classified of the 4 categories of participants are shown in Table 1. For changes
into 3 types: cerebral infarction, cerebral hemorrhage, and sub- in proteinuria, 77.4% had no proteinuria, whereas 7.14% had remit-
arachnoid hemorrhage. Subarachnoid hemorrhage was usually based tent proteinuria, 10.60% had incident proteinuria, and 4.81% had
on aneurysm or vascular malformation instead of chronic stroke persistent proteinuria. Differences in age, sex, BMI, physical activ-
risk factors. Considering the small sample size (n = 31) and differ- ity, cholesterol concentration, eGFR, blood pressure, concomitant
ent mechanism, we did not include the subarachnoid hemorrhage morbidities (diabetes or dyslipidemia), and medicine treatment were
group in the subgroup analysis. The diagnostic criteria were consis- found between proteinuria change groups (P < .05) in both 2006 and
tent across all participating hospitals and all stroke outcomes were 2008. No significant differences in participants who smoke or taking
validated by the Data Safety Monitoring Board and the Arbitration antilipidemic agents were observed between groups. Characteristics
Committee for Clinical Outcomes. The outcome information was of participants included and excluded are showed in Table S1.
further confirmed by checking discharge summaries and the medical During a median follow-­u p of 6.89 (6.56, 7.19) years, we
records from medical insurance. For analysis of stroke-­specific data, identified 1197 subjects with incident strokes. The incident
patients who ended up with unrelated causes were considered as stroke rate was 6.32 events per 1000 person-­y ears for subjects
censored. with no proteinuria, 8.76 for remittent proteinuria, 10.96 for in-
cident proteinuria, and 13.87 for persistent proteinuria. We ad-
ditionally analyzed the association between stroke and different
2.5 | Statistical analyses
proteinuria level. The result is shown in Table 2. The relationship
Continuous variables were described as means ± standard deviation (SD) between stroke and proteinuria is not consistent from 2006 to
and were compared using ANOVA or Kruskal-­Wallis test. Categorical var- 2008. Table 3 shows the adjusted HRs of incident stroke and
iables were described as counts (percentages) and were compared using subtypes associated with changes in proteinuria exposure.
the chi-­square test or the Fisher exact test. Incidence rates of stroke were Compared with subjects without proteinuria, HRs for stroke
calculated per 1000 person-­years of follow-­up. Participants contributed were 1.63 (95% CI, 1.38-­1 .92) for incident proteinuria, and 2.04
person-­time of follow-­up from the date of return of the 2-­year survey (95% CI, 1.65-­2 .53) for persistent proteinuria (Model 1), attenu-
questionnaire to either the date of stroke onset, death, or end of follow- ated but still significant after fully adjustment. The same trends
­up. Cox proportional-­hazards regression was used to calculate the hazard were found in the association between changes in proteinuria
4       WANG et al.

TA B L E   1   Characteristics in 2006 and 2008 according to the change in proteinuria from 2006 to 2008 in participants with hypertension

Change of proteinuria

Variable Total No proteinuria Remittent proteinuria Incident proteinuria Persistent proteinuria P value

No. of participants 24 300 20 665 593 2430 612

Age, years 55.03 ± 10.88 54.89 ± 10.86 55.78 ± 11.10 55.80 ± 10.91 55.97 ± 11.06 <.0001
Sex, male, n (%) 20 010 (82.35) 16 295 (81.90) 491 (82.80) 2084 (85.76) 510 (83.33) <.0001
Baseline information in 2006
Current smoker, 8686 (35.74) 7503 (36.31) 226 (38.11) 784 (32.26) 173 (28.27) <.0001
n (%)
Current alcohol, 9602 (39.51) 8370 (40.50) 228 (38.45) 818 (33.66) 186 (30.39) <.0001
n (%)
Active physical 5024 (20.67) 4401 (21.30) 166 (27.99) 373 (15.35) 84 (13.73) <.0001
activity, n (%)
Antihypertension 6360 (26.17) 5431 (26.28) 193 (32.55) 564 (23.21) 172 (28.10) <.0001
agents, n (%)
Antidiabetes 891 (3.67) 683 (3.31) 38 (6.41) 118 (4.86) 52 (8.50) <.0001
mellitus agents,
n (%)
Antilipidemic 366 (1.51) 296 (1.43) 10 (1.69) 44 (1.81) 16 (2.61) .0582
agents, n (%)
Diabetes mellitus, 3151 (12.97) 2365 (11.44) 152 (25.63) 433 (17.82) 201 (32.84) <.0001
n (%)
Dyslipidemia, 10 444 (42.98) 8710 (42.15) 313 (52.78) 1096 (45.10) 325 (53.10) <.0001
n (%)
Body mass index, 26.13 ± 3.47 26.06 ± 3.44 26.60 ± 3.54 26.41 ± 3.48 27.05 ± 4.02 <.0001
Systolic blood 147.50 ± 17.54 146.80 ± 17.15 152.79 ± 21.19 150.42 ± 18.33 154.41 ± 19.63 <.0001
pressure, mm
Diastolic blood 92.40 ± 10.43 92.08 ± 10.17 95.00 ± 12.39 93.62 ± 11.23 95.60 ± 12.38 <.0001
pressure, mm
Fasting blood 5.68 ± 1.83 5.60 ± 1.68 6.43 ± 2.53 5.93 ± 2.25 6.79 ± 2.98 <.0001
glucose, mmol/L
Total cholesterol, 5.16 ± 1.10 5.16 ± 1.07 5.32 ± 1.19 5.07 ± 1.21 5.24 ± 1.36 .0001
Triglycerides, 1.84 ± 1.48 1.80 ± 1.44 2.05 ± 1.58 2.00 ± 1.59 2.42 ± 1.92 <.0001
Low-­density 2.43 ± 0.99 2.45 ± 0.94 2.57 ± 1.08 2.33 ± 1.23 2.29 ± 1.23 <.0001
High-­density 1.54 ± 0.41 1.54 ± 0.41 1.54 ± 0.46 1.58 ± 0.44 1.57 ± 0.42 <.0001
Estimated 79.75 (25.42) 80.23 (23.16) 80.01 (22.44) 77.57 (40.45) 72.13 (22.16) <.0001
filtration rate,
mL/min/1.73 m²
2008-­2009 follow-­up
Current smoker, 8726 (35.91) 7410 (35.86) 209 (35.24) 890 (36.63) 217 (35.46) .8653
n (%)
Current alcohol, 9109 (37.49) 7827 (37.88) 209 (35.24) 872 (35.88) 201 (32.84) .0123
n (%)

WANG et al. |

TA B L E   1   (Continued)

Change of proteinuria

Variable Total No proteinuria Remittent proteinuria Incident proteinuria Persistent proteinuria P value

Active physical 5513 (22.69) 4869 (23.56) 135 (22.77) 396 (16.30) 113 (18.46) <.0001
activity, n (%)
Antihypertension 4855 (19.98) 3988 (19.30) 159 (26.81) 523 (21.52) 185 (30.23) <.0001
agents, n (%)
Antidiabetes 1192 (4.91) 880 (4.26) 61 (10.29) 169 (6.95) 82 (13.40) <.0001
mellitus agents,
n (%)
Antilipidemic 366 (1.51) 296 (1.43) 10 (1.69) 44 (1.81) 16 (2.61) .0582
agents, n (%)
Diabetes mellitus, 1812 (7.46) 1351 (6.54) 83 (14.00) 257 (10.58) 121 (19.77) <.0001
n (%)
Dyslipidemia, 2493 (10.26) 2117 (10.24) 70 (11.80) 221 (9.09) 85 (13.89) .0031
n (%)
Body mass index, 25.87 ± 3.45 25.83 ± 3.43 26.19 ± 3.44 25.95 ± 3.52 26.56 ± 3.91 <.0001
Systolic blood 144.52 ± 20.09 143.59 ± 19.58 148.35 ± 20.94 149.13 ± 22.32 154.06 (21.44) <.0001
pressure, mm
Diastolic blood 90.70 ± 11.87 90.31 ± 11.61 91.43 ± 12.32 92.83 ± 13.09 94.62 ± 13.31 <.0001
pressure, mm
Fasting blood 5.95 ± 2.05 5.85 ± 1.90 6.41 ± 2.20 6.37 ± 2.73 7.15 ± 2.96 <.0001
glucose, mmol/L
Total cholesterol, 5.17 ± 1.56 5.16 ± 1.56 5.30 ± 1.22 5.23 ± 1.73 5.42 ± 1.29 <.0001
Triglycerides, 1.79 ± 1.97 1.75 ± 1.92 2.04 ± 1.85 1.88 ± 2.10 2.38 ± 2.84 <.0001
Low-­density 2.71 ± 1.36 2.70 ± 1.40 2.83 ± 0.98 2.79 ± 1.73 2.80 ± 0.96 <.0001
High-­density 1.52 ± 0.62 1.51 ± 0.49 1.64 ± 1.77 1.51 ± 0.95 1.49 ± 0.48 .0004
Estimated 82.01 ± 21.53 82.90 ± 21.26 82.14 ± 22.89 76.43 ± 21.45 73.96 ± 24.32 <.0001
filtration rate,
mL/min/1.73 m²
All-­t ype stroke, 1197 (4.93) 920 (4.45) 42 (7.08) 183 (7.53) 52 (8.50) <.0001
n (%)
Ischemic stroke 1003 (4.13) 779 (3.77) 36 (6.07) 143 (5.88) 45 (7.35) <.0001
Intracerebral 203 (0.84) 151 (0.73) 6 (1.01) 38 (1.56) 8 (1.31) .0002
Subarachnoid 31 (0.13) 23 (0.11) 0 (0.00) 7 (0.29) 1 (0.16) .1040

and stroke subtypes (ischemic stroke and hemorrhagic stroke). hemorrhage. Sensitivity analysis yielded the same pattern of re-
Notably, each degree of proteinuria decline was associated with sults in this entire test.
a significant 12% reduction of stroke risk (HR 0.88, 95% CI, 0.82-­ Figure 2 shows the Kaplan-­Meier cumulative risk for stroke and
0.93), independent from age, sex, and other potential risk fac- subtypes within groups defined by changes in proteinuria. Participants
tors. Findings were consistent for stroke subtypes, with a 11% in the persistent proteinuria group experienced a higher risk than par-
decrease (HR 0.89, 95% CI, 0.83-­0 .96) for ischemic stroke and ticipants in the other groups during the 6.89-­year follow-­up period for
a 19% decrease (HR 0.81, 95% CI, 0.70-­0 .94) for intracerebral total stroke events (log-­rank test, P < .01) (Figure 2A), ischemic stroke
6       WANG et al.

TA B L E   2   Stroke incidence rate and hazard ratios for stroke risk by proteinuria in participants with hypertension in 2006 and 2008

Proteinuria Stroke Ischemic stroke Intracerebral hemorrhage

2006-­2007 follow-­up − Ref Ref Ref

Trace 1.26 (1.04-­1.54) 1.25 (1.01-­1.54) 1.57 (1.01-­2.43)
+ 1.01 (0.73-­1.39) 1.08 (0.77-­1.51) 0.63 (0.23-­1.71)
++ 1.25 (0.84-­1.86) 1.36 (0.90-­2.05) 0.96 (0.30-­3.03)
+++ 2.51 (1.71-­3.70) 2.19 (1.40-­3.43) 4.01 (1.86-­8.63)
2008-­2009 follow-­up − Ref Ref Ref
Tracea 1.57 (1.18-­2.09) 1.72 (1.27-­2.31) 1.29 (0.60-­2.78)
+ 1.48 (1.25-­1.76) 1.37 (1.13-­1.66) 1.71 (1.15-­2.54)
++ 1.15 (0.84-­1.58) 1.13 (0.80-­1.59) 1.58 (0.82-­3.04)
+++ 1.57 (1.05-­2.36) 1.49 (0.95-­2.34) 1.77 (0.71-­4.39)
Adjusted for age, gender, current smoker, current alcohol user, physical activity, body mass index, diabetes mellitus, dyslipidemia, total cholesterol,
high-­density lipoprotein, systolic blood pressure, fasting blood glucose, and eGFR.

TA B L E   3   Stroke incidence rate and hazard ratios for stroke risk by changes in proteinuria in participants with hypertension

Changes in proteinuria

Per degree
No proteinuria Remittent proteinuria Incident proteinuria Persistent proteinuria decreasea

All-­t ype stroke

Incidence rate, per 6.32 8.76 10.96 13.87
1000 person-­years
Model 1 Reference 1.37 (1.11-­1.69) 1.63 (1.38-­1.92) 2.04 (1.65-­2.53) 0.82 (0.79-­0.90)
Model 2 Reference 1.36 (1.10-­1.68) 1.60 (1.35-­1.89) 1.99 (1.61-­2.47) 0.85 (0.79-­0.90)
Model 3 Reference 1.20 (0.87-­1.64) 1.41 (1.05-­1.77) 1.49 (1.25-­1.89) 0.88 (0.82-­0.93)
Sensitive analysisb Reference 1.21 (0.88-­1.69) 1.40 (1.04-­1.80) 1.51 (1.27-­1.88) 0.92 (0.86-­0.95)
Ischemic stroke
Incidence rate, per 5.29 7.53 8.93 11.39
1000 person-­years
Model 1 Reference 1.40 (1.12-­1.76) 1.60 (1.34-­1.92) 2.02 (1.60-­2.56) 0.86 (0.80-­0.92)
Model 2 Reference 1.39 (1.11-­1.74) 1.57 (1.31-­1.89) 1.97 (1.55-­2.49) 0.86 (0.80-­0.93)
Model 3 Reference 1.19 (0.84-­1.67) 1.36 (1.12-­1.65) 1.41 (1.03-­1.94) 0.89 (0.83-­0.96)
Sensitive analysisb Reference 1.20 (0.85-­1.69) 1.37 (1.13-­1.67) 1.42 (1.03-­1.96) 0.95 (0.88-­0.97)
Intracerebral hemorrhage
Incidence rate, per 1.04 1.33 1.83 2.64
1000 person-­years
Model 1 Reference 1.29 (0.76-­2.19) 1.65 (1.11-­2.46) 2.38 (1.46-­3.87) 0.80 (0.69-­0.92)
Model 2 Reference 1.29 (0.76-­2.18) 1.65 (1.11-­2.47) 2.40 (1.47-­3.91) 0.79 (0.68-­0.92)
Model 3 Reference 1.15 (0.50-­2.61) 1.99 (1.35-­2.93) 2.49 (1.71-­3.10) 0.81 (0.70-­0.94)
Sensitive analysis Reference 1.15 (0.50-­2.62) 2.04 (1.38-­3.00) 2.34 (1.61-­2.93) 0.87 (0.75-­1.01)

Model 1: adjusted for age and gender.

Model 2: adjusted for age, gender, current smoker, current alcohol user, physical activity, and body mass index.
Model 3: adjusted for variables in model 2 plus diabetes mellitus, dyslipidemia, total cholesterol, high-­density lipoprotein, systolic blood pressure,
fasting blood glucose, and eGFR.
Adjusted for proteinuria at baseline survey in addition to model 1, 2 and 3, respectively.
Adjusted for model 3 and further excluded individuals with an estimated glomerular filtration rate < 30 mL/min per 1.73 m2 in 2008.
WANG et al. |

F I G U R E   2   Kaplan-­Meier curve of (A)

all-­t ype stroke incidence rate; (B) ischemic
stroke incidence rate; (C) intracerebral
hemorrhage incidence rate by changes
in proteinuria in participants with
8       WANG et al.

(log-­rank test, P < .01) (Figure 2B), and hemorrhagic stroke (log-­rank Chronic kidney disease was highly prevalent among patients with
test, P < .01) (Figure 2C). type 2 diabetes mellitus.24 According to previous studies, proteinuria
No interactions of proteinuria with diabetes mellitus (P = .14) or was believed to be associated with diabetes mellitus. 41% increased
sex (P = .25) were found. risks for persistent proteinuria development over 20 years were
found among 1304 participants with diabetes in the DCCT/EDIC
study.25 Elevated insulin levels are associated with increased vascular
4 |  D I S CU S S I O N inflammation and, hence, the connection with proteinuria.26 It is im-
portant to emphasize that this result showed there was no evidence
In our study, the presence of proteinuria as detected by urine dip- of effect modification by diabetes mellitus. Changes in proteinuria
stick screening independently predicted an increased risk for incident are associated with stroke in the hypertensive population ignoring
stroke during median 6.89 years of follow-­up. We observed significant diabetes status. As corroboration, results from other studies also
associations of changes in proteinuria and the incidence of stroke and demonstrated that any degree of proteinuria is a risk factor for car-
subtypes in the Chinese hypertensive population. This relationship diovascular events in individuals with or without diabetes mellitus.27
persisted independently of other known major risk factors, including Screening for proteinuria to identify people at high risk of stroke is
diabetes mellitus, dyslipidemia, and obesity. No interactions of protein- essential in the hypertensive population regardless of the diabetes
uria with sex contribute to this relationship. Compared with changes status.
in proteinuria, proteinuria levels in different years are not consistent. The mechanism linking proteinuria to stroke risk remains ob-
This result indicated that compared with single measurement, changes scure. The Steno hypothesis suggested that proteinuria not only
in proteinuria might be a more stable risk factor for stroke. reflects localized renal damage but also is an independent marker
Proteinuria has acted as a risk multiplier of cardiovascular risk of systemic vascular endothelial dysfunction or microvascular dis-
and mortality across all stages of kidney disease.17 Assessing pro- ease.28 Proteinuria is associated with several inflammatory and
teinuria using the simple urine dipstick test could be a convenient thrombogenic factors, which play important roles in the devel-
way to reflect an individual’s systematic vasculature permeability opment of atherosclerosis and stroke.21 Otherwise, proteinuria
and susceptibility to target organ damage. Previous studies have is associated with a high prevalence of traditional cardiovascular
reported that baseline proteinuria was associated with a higher risk factors, such as hypertension and diabetes. Proteinuria may be
stroke risk, with the risk greatest for ischemic stroke. A recent a sign of shared risk factors for vascular disease and not a direct
study from the China Stroke Primary Prevention Trial also found cause of new strokes. Urine protein change over time can be an
that baseline proteinuria as measured by dipstick was an indepen- indicator for the severity of those stroke risk factors. In the hyper-
dent risk factor for first incident stroke and ischemic stroke.20 A tensive population, proteinuria and hypertension have reciprocal
meta-­analysis of cohort studies or randomized controlled trials also effects.7 Increased proteinuria in hypertensive patients is consid-
concluded that increasing albuminuria increased the risk of stroke. ered an early manifestation of generalized vascular impairment,
Findings from the current study extend those from previous stud- which is regarded to be an important marker of the risk of renal
ies. Our study shows that the presence of proteinuria is associated and cardiovascular events.29 According to these findings, we should
with a substantially higher risk of incident stroke compared with pay more attention to proteinuria, especially in the hypertensive
an absence of proteinuria when measured either at baseline or at population. Hypertension and proteinuria are both clinical mani-
follow-­up visits. As proteinuria often changes dynamically, showing festation of renal disease, both also playing important roles in car-
regression or progression,9 changes in proteinuria are more likely diovascular and cerebrovascular events. Whether proteinuria is a
to reflect the risk of stroke, compared with proteinuria collected therapeutic target for stroke warrants clinical attention and further
at a single time point in the hypertensive population. Previous investigations.
study showed a 10% proteinuria reduction corresponded to 29% Strengths of our study were the ability to consider dynamic as-
reduction of stroke in people with hypertension and/or diabetes.22 pects of proteinuria using the biennial assessment of proteinuria to
Current result showed that each degree of proteinuria decline was estimate the risk of stroke, the prospective design, use of a large
associated with a significant 12% reduction of future stroke risk. cohort, and long follow-­up period. Our results should be interpreted
This reinforces the concept that stroke risk is reduced when pro- in the context of some limitations. First, awareness of the limita-
teinuria is diminished, independent of other traditional cardiovas- tions of the dipstick test is necessary. The dipstick method could
cular risk factors. overlook proteinuria.14 A timed 24 hours urine collection or albu-
Our study investigated this association by distinguishing between min:creatinine ratio might be more precise in measuring proteinuria,
ischemic and hemorrhagic stroke, as the patterns of results were con- although dipstick has a cost advantage.30 Second, we have adjusted
sistent with total stroke. The effect size of changes in proteinuria for confounding variables when evaluating the proteinuria recorded in
risks of stroke subtypes including ischemic stroke and hemorrhagic different years, but there were still some factors we failed to take
stroke was similar. In accordance with this result, previous studies into consideration. Last, unlike a randomized clinical trial, protein-
also showed albuminuria was significant association with both stroke uria could be associated with substantial morbidity and mortality in
types.23 the population we excluded, which might be especially vulnerable
WANG et al. |

to bias due to attrition. Additionally, proteinuria may be a sign of 9. Brantsma AH, Bakker SJ, de Zeeuw D, de Jong PE, Gansevoort
shared risk factors for vascular disease and chronic kidney disease. RTGroup PS. Extended prognostic value of urinary albumin excretion
for cardiovascular events. J Am Soc Nephrol. 2008;19:1785‐1791.
The association between proteinuria and stroke may contributed by
10. Zhang Q, Zhou Y, Gao X, et al. Ideal cardiovascular health metrics
multiple effect. Specific diseases such as chronic glomerulonephritis and the risks of ischemic and intracerebral hemorrhagic stroke.
cannot be simply ruled out. Urine protein change over time can be Stroke. 2013;44:2451‐2456.
an indicator for the severity of those stroke risk factors. Because 11. Wang A, Chen S, Wang C, et al. Resting heart rate and risk of cardio-
vascular diseases and all-­cause death: the Kailuan study. PLoS ONE.
observational study is unable to determine causality in its findings,
whether changes in proteinuria are a risk factor or treatment target 12. Peng M, Wu S, Jiang X, Jin C, Zhang W, Kailuan Cardiovascular
needs more exploration. Survey G. Long-­term alcohol consumption is an independent risk
factor of hypertension development in northern China: evidence
from Kailuan study. J Hypertens. 2013;31:2342‐2347.
13. Liu X, Cui L, Wang A, et al. Cumulative exposure to ideal cardio-
5 | CO N C LU S I O N S vascular health and incident diabetes in a Chinese population: the
Kailuan study. J Am Heart Assoc. 2016;5:e004132.
In conclusion, we emphasized that changes in proteinuria, particu- 14. White SL, Yu R, Craig JC, Polkinghorne KR, Atkins RC, Chadban SJ.
Diagnostic accuracy of urine dipsticks for detection of albuminuria
larly persistent proteinuria, were significantly associated with inci-
in the general community. Am J Kidney Dis. 2011;58:19‐28.
dent stroke in participants with hypertension. Our study provides 15. Li Z, Wang A, Cai J, et al. Impact of proteinuria and glomerular filtra-
valuable information to public health and medical professionals by tion rate on risk of ischaemic and intracerebral hemorrhagic stroke:
examining a dynamic relationship between changes in proteinuria a result from the Kailuan study. Eur J Neurol. 2015;22:355‐360.
16. Stroke–1989. Recommendations on stroke prevention, diagnosis,
and incident stroke in the hypertensive population.
and therapy. Report of the WHO Task Force on Stroke and Other
Cerebrovascular Disorders. Stroke. 1989;20:1407‐1431.
17. Gansevoort RT, Correa-Rotter R, Hemmelgarn BR, et al. Chronic
C O N FL I C T O F I N T E R E S T kidney disease and cardiovascular risk: epidemiology, mechanisms,
and prevention. Lancet. 2013;382:339‐352.
None of the authors have competing financial interests to declare.
18. Eckardt KU, Coresh J, Devuyst O, et al. Evolving importance of
kidney disease: from subspecialty to global health burden. Lancet.
ORCID 19. Zhang C, Wang X, He M, et al. Proteinuria is an independent risk
factor for first incident stroke in adults under treatment for hyper-
Yongjun Wang
tension in China. J Am Heart Assoc. 2015;4:e002639.
Yilong Wang 20. Zhang Y, Zhang T, Zhang C, et al. Identification of reciprocal cau-
sality between non-­alcoholic fatty liver disease and metabolic syn-
drome by a simplified Bayesian network in a Chinese population.
BMJ Open. 2015;5:e008204.
21. Masson P, Webster AC, Hong M, Turner R, Lindley RI, Craig JC.
1. Li H, Meng Q, Sun X, Salter A, Briggs NE, Hiller JE. Prevalence, Chronic kidney disease and the risk of stroke: a systematic review
awareness, treatment, and control of hypertension in rural and meta-­analysis. Nephrol Dial Transplant. 2015;30:1162‐1169.
China: results from Shandong Province. J Hypertens. 2010;28: 22. Savarese G, Dei Cas A, Rosano G, et al. Reduction of albumin urinary
432‐438. excretion is associated with reduced cardiovascular events in hyper-
2. Dries DL, Exner DV, Domanski MJ, Greenberg B, Stevenson LW. tensive and/or diabetic patients. A meta-­regression analysis of 32 ran-
The prognostic implications of renal insufficiency in asymptomatic domized trials. Int J Cardiol. 2014;172:403‐410.
and symptomatic patients with left ventricular systolic dysfunction. 23. Mahmoodi BK, Yatsuya H, Matsushita K, et al. Association of kidney dis-
J Am Coll Cardiol. 2000;35:681‐689. ease measures with ischemic versus hemorrhagic strokes: pooled analyses
3. Best PJ, Lennon R, Ting HH, et al. The impact of renal insufficiency of 4 prospective community-­based cohorts. Stroke. 2014;45:1925‐1931.
on clinical outcomes in patients undergoing percutaneous coronary 24. Low SK, Sum CF, Yeoh LY, et al. Prevalence of chronic kidney disease
interventions. J Am Coll Cardiol. 2002;39:1113‐1119. in adults with type 2 diabetes mellitus. Ann Acad Med Singapore.
4. Manjunath G, Tighiouart H, Ibrahim H, et al. Level of kidney func- 2015;44:164‐171.
tion as a risk factor for atherosclerotic cardiovascular outcomes in 25. Boger CA, Chen MH, Tin A, et al. CUBN is a gene locus for albumin-
the community. J Am Coll Cardiol. 2003;41:47‐55. uria. J Am Soc Nephrol. 2011;22:555‐570.
5. Ovbiagele B. Microalbuminuria: risk factor and potential therapeu- 26. Bakris GL, Molitch M. Microalbuminuria as a risk predictor in diabe-
tic target for stroke? J Neurol Sci. 2008;271:21‐28. tes: the continuing saga. Diabetes Care. 2014;37:867‐875.
6. Lee M, Saver JL, Chang KH, Liao HW, Chang SC, Ovbiagele B. 27. Gerstein HC, Mann JF, Yi Q, et al. Albuminuria and risk of cardio-
Impact of microalbuminuria on incident stroke: a meta-­analysis. vascular events, death, and heart failure in diabetic and nondiabetic
Stroke. 2010;41:2625‐2631. individuals. JAMA. 2001;286:421‐426.
7. Tenekecioglu E, Yilmaz M, Yontar OC, et al. Microalbuminuria in un- 28. Deckert T, Feldt-Rasmussen B, Borch-Johnsen K, Jensen T, Kofoed-
treated prehypertension and hypertension without diabetes. Int J Enevoldsen A. Albuminuria reflects widespread vascular damage.
Clin Exp Med. 2014;7:3420‐3429. The Steno hypothesis. Diabetologia. 1989;32:219‐226.
8. Guan T, Ma J, Li M, et al. Rapid transitions in the epidemiology of 29. Leoncini G, Sacchi G, Viazzi F, et al. Microalbuminuria identifies
stroke and its risk factors in China from 2002 to 2013. Neurology. overall cardiovascular risk in essential hypertension: an artificial
2017;89:53‐61. neural network-­based approach. J Hypertens. 2002;20:1315‐1321.
10       WANG et al.

3 0. Levin A, Stevens PE. Summary of KDIGO 2012 CKD Guideline: be-

hind the scenes, need for guidance, and a framework for moving How to cite this article: Wang A, Dai L, Su Z, et al.
forward. Kidney Int. 2014;85:49‐61. Proteinuria and risk of stroke in patients with hypertension:
The Kailuan cohort study. J Clin Hypertens. 2018;00:1–10.


Additional Supporting Information may be found online in the

supporting information tab for this article.