Abstract
Aims
To examine the relationship between systolic blood pressure (SBP) variability and the risk of microvascular complications in a non-elderly diabetic population.
Methods
This is a retrospective cohort study of individuals aged ≤60 years treated for diabetes in 2003 in the US Department of Veterans Affairs healthcare system. Individuals were followed for five years for any new diagnosis of diabetic nephropathy, retinopathy, or neuropathy. In each year of follow-up, individuals were classified into quartiles based on their SBP variability.
Results
We identified 208,338 patients with diabetes without diabetic nephropathy, retinopathy, or neuropathy at baseline. Compared to individuals with the least SBP variability (Quartile 1), those with most variability (Quartile 4) had 81% (OR = 1.81; 95% CI, 1.72–1.91), 17% (OR = 1.17; 95% CI, 1.13–1.21), 30% (OR = 1.30; 95% CI, 1.25–1.35), and 19% (OR = 1.19; 95% CI, 1.15–1.23) higher incidence of nephropathy, retinopathy, neuropathy, and any complication, respectively, after adjusting for mean SBP, demographic and clinical factors.
Conclusions
We found a significant graded relationship between SBP variability and the incidence of each complication and of any combined endpoint. This is the first study showing a significant association between SBP variability and the risk of diabetic retinopathy and neuropathy.
Keywords: Blood pressure variability, Microvascular complications, Neuropathy, Retinopathy, Nephropathy
1. Introduction
Blood pressure control is important for individuals with diabetes to reduce the risk of cardiovascular disease (CVD) and other diabetic complications (American Diabetes Association, 2015). Trial evidence shows that intensive blood pressure control reduces both microvascular and macrovascular complications (Curb, Pressel, Cutler, et al., 1996; Tuomilehto, Rastenyte, Birkenhager, et al., 1999; UK Prospective Diabetes Study Group, 1998). In the past few years, studies have shown that blood pressure variability (BPV) is an independent predictor of coronary heart disease, stroke, cognitive dysfunction, and all-cause mortality (Epstein, Lane, Farlow, et al., 2013; Muntner, Whittle, Lynch, et al., 2015; Muntner et al., 2011; Rothwell, Howard, Dolan, et al., 2010a) and that BPV may directly cause end organ damage (Su, 2006). In diabetic patients, several studies (Kilpatrick, Rigby, & Atkin, 2010; Okada, Fukui, Tanaka, et al., 2012; Okada, Fukui, Tanaka, et al., 2013; Okada, Matsumoto, Nagaoka, & Nakao, 2012; Takao, Matsuyama, Yanagisawa, Kikuchi, & Kawazu, 2014) have consistently shown that excessive BPV is an independent predictor of nephropathy and its precursors such as micro- (Noshad, Mousavizadeh, Mozafari, Nakhjavani, & Esteghamati, 2014) and macroalbuminuria (Ushigome, Fukui, Hamaguchi, et al., 2011). However, the relationship between BPV and retinopathy or neuropathy is still uncertain. Our objective in this study is to examine the relationship between visit-to-visit blood pressure variability and diabetic microvascular complications using a large administrative database of patients with diabetes. Our hypothesis is that excessive BPV is associated with increased risk of developing these conditions.
2. Material and methods
2.1. Design overview
We used a retrospective cohort design to follow patients with diabetes for five years and to examine whether their SBP variability in a given year was associated with an increased risk of a microvascular complication in the next year. A person-year data set was constructed in which patients without previous evidence of a microvascular complication were followed each year until they experienced a complication or death.
2.2. Setting and participants
We used a cohort of diabetic patients treated in the US Department of Veterans Affairs (VA) healthcare system in the fiscal year 2003 (October 1, 2002–September 30, 2003; all years henceforth are fiscal years). Inpatient and outpatient records (Veterans Affairs Information Resource Center, 2005a, 2005b) were used to identify diabetic patients who had one or more prescriptions of diabetes medications filled in 2003 or had one or more hospitalizations or two or more outpatient visits with a diagnostic code for diabetes (ICD-9-CM 250.xx) in 2002–2003. A previous study showed that this method had a 93% sensitivity and 97% specificity (Miller, Safford, & Pogach, 2004) against self-reported diabetes.
Because Medicare data were not available, patients who turned 65 years of age before the end of the study period (September 30, 2008) were excluded to mitigate the potential bias in ascertaining study outcomes occurring outside the VA healthcare system.
2.3. Visit-to-visit variability of systolic blood pressure
We obtained all outpatient BP measures recorded in 2003–2007 for our study cohort from the VA Corporate Data Warehouse. All measures taken on days patients were hospitalized or made an emergency room visit were dropped. We then selected measures of BP as valid if they were between 50 and 300 mm Hg for systolic blood pressure (SBP) and 30–180 mm Hg for diastolic blood pressure (DBP). Multiple BP measures taken during the same day for the same patient were averaged as the measure of the day and only patients with three or more BP measurements in the same year were included in the study cohort. For patients with more than 12 measures in a year, we randomly selected 12 measures for that year (Rothwell et al., 2010a). We computed within-subject means and standard deviations of SBP for each year and used them to compute the SBP variability independent of the mean (VIM) by following the method used in previous studies (Rothwell, Howard, Dolan, et al., 2010b; Rothwell et al., 2010a; Schutte, Thijs, Liu, et al., 2012).
2.4. Outcomes and follow-up
Four main outcomes were identified using ICD-9-CM codes reported in both inpatient and outpatient data for 2000–2008: retinopathy, 250.5x, 366.41, and 362.0x; nephropathy, 250.4x and 583.81; neuropathy, 250.6x, 354.xx, 337.1 and 357.2, or any of the three complications. We searched inpatient and outpatient records for four years (2000–2003) before the baseline (October 1, 2003) to identify individuals with pre-existing complications and excluded them from the study cohort. The incidence date of a condition was determined as the earliest date the condition was recorded.
We constructed a separate longitudinal data set for each outcome, in which the time interval was a year (to allow for adequate time for multiple BP measures), each individual was observed for the outcome event once every year for five years (2004–2008) or until death, and the SBP variability and other covariates were derived from one year prior (2003–2007). A patient’s record was repeated once each year in the analytic file as long as the patient had three or more BP measures in a year, had not been diagnosed with the condition during the year or before, and did not die until the end of the next year. A patient could thus be included in our person-year data up to five times.
2.5. Statistical analysis
We conducted bivariate and multivariable analyses to compare the differences in risks of diabetic microvascular complications between VIM quartiles. Multivariable models included adjustments for year, demographic factors (age, sex, race/ethnicity, and marital status), clinical factors that indicate diabetes severity or are known to affect BP or vascular health (A1c, diabetes duration >5 years, diabetes medication treatment type, BMI, antihypertensive medication treatment, and statin use), and mean SBP. Demographic factors were measured at baseline. All clinical factors and mean SBP were measured using data recorded during one year before outcomes. Because access to care and frequency of healthcare encounters can affect timing of disease diagnosis, we also obtained copayment status for VA healthcare services, geographic distance to the nearest VA outpatient clinic from patients’ residence, and the number of visits and hospitalizations in the previous year. These variables were excluded in final models, because otherwise identical models with or without these variables showed virtually identical estimates for VIM quartiles.
Diabetes treatment was defined based on diabetic medications filled for 30 days or more during a given year as follows: no use, insulin only, oral medication only, and both insulin and oral medication use. Antihypertensive medication use or statin use was indicated when any antihypertensive medication or statin prescription was filled for 30 days or longer in a year.
For each outcome, we estimated a separate but identically specified model that adjusted for all covariates including mean SBP. Each model was analyzed using a random-intercept logistic regression to account for the repeated nature of data.
We conducted sensitivity analyses to compare the outcomes between VIM quartiles by stratifying the cohort into individuals with high (≥130/80 mm Hg) and low (<130/80 mm Hg) BP according to recommended BP targets for younger diabetic patients in the American Diabetes Association (ADA) Standards of Medical Care in Diabetes (American Diabetes Association, 2014) and into individuals with 3–6 BP measures and those with 7 or more measures a year. Six was the median number of BP measures per year in our study cohort. The latter stratified analysis was intended to address the concern that the association between SBP variability and outcomes may be confounded by the number of BP measures.
We further limited our sample to a subset of normotensive patients (BP < 130/80 mm Hg) and stratified them into antihypertensive medication users and non-users to examine whether they are different in their SBP variability-outcome associations.
In this study, all outcomes were identified by any single code during the study period. We conducted another sensitivity analysis to test whether our results would change substantially if outcomes were identified by different methods. We considered two alternative methods of outcome identification, the first one requiring only an inpatient or two outpatient ICD-9-CM codes and the second, two codes, any time during the study period (Lee, Shields, Vogeli, et al., 2007; Miller et al., 2004; Sohn, Budiman-Mak, Stuck, Siddiqui, & Lee, 2010).
This study was approved by our institutional review board for the research of human subjects.
3. Results
The study cohort included 208,338 patients, followed for an average of 3.5 ± 1.4 years. Patients were 53.7 ± 5.2 years old during follow-up, mostly male (95.4%), and married in 46.5% of person-years (Table 1). Fifty-three percent were non-Hispanic white, 19.8% were non-Hispanic black, and 5.3% were Hispanic. Their mean BP was 134.5/76.4 ± 13.8/8.3 mm Hg and 88.7% of person-years were treated with antihypertensive medications.
Table 1.
Demographic and clinical characteristics of the study sample by systolic blood pressure variability independent of the mean quartiles.a
| Demographic and clinical factors | All | SBP variability quartiles
|
P-Value | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 (Least) | 2 | 3 | 4 (Most) | ||||||||
| N | 726,930 | (100.0) | 181,734 | (100.0) | 181,732 | (100.0) | 181,733 | (100.0) | 181,731 | (100.0) | |
| Demographic factors | |||||||||||
| Age, mean (SD) | 53.7 | (5.2) | 53.2 | (5.6) | 53.5 | (5.3) | 53.8 | (5.0) | 54.1 | (4.7) | <0.001 |
| Male sex | 693,612 | (95.4) | 173,729 | (95.6) | 173,051 | (95.2) | 173,155 | (95.3) | 173,678 | (95.6) | <0.001 |
| Race/ethnicity | |||||||||||
| NH White | 382,917 | (52.7) | 92,323 | (50.8) | 96,164 | (52.9) | 97,732 | (53.8) | 96,698 | (53.2) | <0.001 |
| NH Black | 143,720 | (19.8) | 31,923 | (17.6) | 34,580 | (19.0) | 36,677 | (20.2) | 40,540 | (22.3) | |
| Hispanic | 38,292 | (5.3) | 10,373 | (5.7) | 9899 | (5.4) | 9216 | (5.1) | 8804 | (4.8) | |
| Other/Unknown | 162,001 | (22.3) | 47,115 | (25.9) | 41,089 | (22.6) | 38,108 | (21.0) | 35,689 | (19.6) | |
| Married | 338,246 | (46.5) | 85,046 | (46.8) | 86,743 | (47.7) | 85,200 | (46.9) | 81,257 | (44.7) | <0.001 |
| Clinical factors | |||||||||||
| Systolic blood pressure | |||||||||||
| Mean (between-subject SD) | 134.5 | (13.8) | 132.9 | (12.6) | 133.2 | (13.1) | 134.4 | 13.8) | 137.5 | (15.2) | <0.001 |
| Within-subject maximum | 151.4 | (18.6) | 140.1 | (13.3) | 146.6 | (14.3) | 153.1 | (15.7) | 165.7 | (10.0) | <0.001 |
| Within-subject standard deviation | 12.7 | (6.1) | 6.0 | (1.9) | 10.2 | (1.1) | 13.8 | (1.4) | 20.8 | (4.8) | <0.001 |
| Diastolic blood pressure | |||||||||||
| Mean (between-subject SD) | 76.4 | (8.3) | 76.6 | (7.9) | 76.2 | (8.0) | 76.2 | (8.3) | 76.7 | (9.0) | <0.001 |
| Within-subject maximum | 86.3 | (10.1) | 83.3 | (8.6) | 85.0 | (9.1) | 86.6 | (9.7) | 90.2 | (11.5) | <0.001 |
| Within-subject standard deviation | 7.6 | (3.4) | 5.7 | (2.7) | 6.8 | (2.7) | 7.8 | (2.8) | 10.1 | (3.8) | <0.001 |
| A1c | |||||||||||
| <7% | 220,492 | (30.3) | 56,137 | (30.9) | 56,043 | (30.8) | 55,240 | (30.4) | 53,072 | (29.2) | <0.001 |
| 7%–9% | 260,952 | (35.9) | 66,431 | (36.6) | 66,309 | (36.5) | 65,648 | (36.1) | 62,564 | (34.4) | |
| >9% | 128,279 | (17.6) | 30,055 | (16.5) | 30,982 | (17.0) | 32,210 | (17.7) | 35,032 | (19.3) | |
| Unknown | 117,207 | (16.1) | 29,111 | (16.0) | 28,398 | (15.6) | 28,635 | (15.8) | 31,063 | (17.1) | |
| Diabetes duration in 2003 >5 y | 197,611 | (27.2) | 42,179 | (23.2) | 47,171 | (26.0) | 51,135 | (28.1) | 57,126 | (31.4) | <0.001 |
| Diabetes treatment | |||||||||||
| None | 69,986 | (9.6) | 18,202 | (10.0) | 17,707 | (9.7) | 16,901 | (9.3) | 17,176 | (9.5) | <0.001 |
| Oral medications alone | 354,007 | (48.7) | 99,569 | (54.8) | 91,047 | (50.1) | 85,467 | (47.0) | 77,924 | (42.9) | |
| Insulin alone | 107,095 | (14.7) | 21,173 | (11.7) | 24,027 | (13.2) | 27,511 | (15.1) | 34,384 | (18.9) | |
| Oral and insulin | 195,842 | (26.9) | 42,790 | (23.5) | 48,951 | (26.9) | 51,854 | (28.5) | 52,247 | (28.7) | |
| Body mass index, kg/m2 | |||||||||||
| 25 | 67,041 | (9.2) | 14,027 | (7.7) | 15,080 | (8.3) | 16,457 | (9.1) | 21,477 | (11.8) | <0.001 |
| 25–29.9 | 186,383 | (25.6) | 47,693 | (26.2) | 46,270 | (25.5) | 45,699 | (25.1) | 46,721 | (25.7) | |
| ≥30 | 462,431 | (63.6) | 116,552 | (64.1) | 117,753 | (64.8) | 117,195 | (64.5) | 110,931 | (61.0) | |
| Unknown | 11,075 | (1.5) | 3462 | (1.9) | 2629 | (1.4) | 2382 | (1.3) | 2602 | (1.4) | |
| Statin use | 378,570 | (52.1) | 97,655 | (53.7) | 96,452 | (53.1) | 94,589 | (52.0) | 89,874 | (49.5) | <0.001 |
| Antihypertensive medication use | 644,582 | (88.7) | 152,469 | (83.9) | 158,737 | (87.3) | 164,112 | (90.3) | 169,264 | (93.1) | <0.001 |
NH = non-Hispanic; SD = standard deviation; A1c = hemoglobin A1c; SBP VIM = systolic blood pressure variability independent of the mean.
All numbers inside parentheses are column percentages unless otherwise indicated; systolic blood pressure (SBP) variability quartiles were based on the SBP variability independent of the mean and value ranges in 2003 were as follows: 1st quartile = 0–0.687; 2nd quartile = 0.687–0.950; 3rd quartile = 0.951–1.224; 4th quartile = 1.225–4.93. The ranges for the other years were similar to the ones for 2003.
Persons with older age, non-Hispanic white or black races, hemoglobin A1c >9%, diabetes duration >5 years, BMI <25 kg/m2, and those who used insulin alone or antihypertensive medications were more likely to be in Quartile 4 (most variable) than Quartile 1 (least variable, P < 0.001).
Mean SBP increased slightly between Quartiles 1 and 4 but mean within-subject SBP standard deviations (SD) increased by 4 mm Hg or more between each successive quartiles (6.0, 10.2, 13.8, and 20.8 mm Hg for Quartiles 1–4; p < 0.001). Mean DBP did not substantially change between quartiles but mean within-subject DBP SDs increased monotonically by 1 mm Hg or more in each higher quartile (5.7, 6.8, 7.8, and 10.1 mm Hg; P < 0.001).
One-year incidence rates were 2.2%, 6.1%, 6.5%, and 11.1%, respectively, for nephropathy, retinopathy, neuropathy, and any complication in 2004 (Table 2). Five-year cumulative incidence rates were 8.5%, 23.8%, 23.3% and 39.8%, respectively.
Table 2.
Number at risk and five-year cumulative incidence of microvascular complications.
| Year | Nephropathy
|
Retinopathy
|
Neuropathy
|
All complications
|
||||
|---|---|---|---|---|---|---|---|---|
| Number | Incidence | Number | Incidence | Number | Incidence | Number | Incidence | |
| 2003 | 143,736 | 2.23 | 112,363 | 6.08 | 117,311 | 6.47 | 87,135 | 11.11 |
| 2004 | 133,381 | 2.14 | 99,074 | 5.49 | 103,333 | 6.55 | 72,010 | 10.68 |
| 2005 | 125,893 | 2.06 | 90,305 | 5.11 | 92,955 | 5.75 | 61,992 | 9.54 |
| 2006 | 118,241 | 1.92 | 81,940 | 5.74 | 83,964 | 5.42 | 53,614 | 9.96 |
| 2007 | 114,878 | 1.87 | 76,324 | 5.78 | 78,899 | 5.26 | 48,086 | 9.88 |
| 5-year | 190,408 | 8.48 | 148,040 | 23.80 | 159,409 | 23.31 | 115,504 | 39.79 |
“Number” indicates the total number of persons who did not have the condition in 2003 and were “at risk” of these complications in 2004–2008; “incidence” is per 100 persons.
Table 3 shows a summary of results from multivariable models (full models are shown in Table A in the online supplement). All odds ratios were statistically significant at P < 0.05 and all VIM quartiles had a significant graded relationship with the risks of all outcomes (P for trend <0.001 in all models). Compared to persons in Quartile 1 (lowest variability), the risk of developing a microvascular complication for persons in Quartile 4 (highest variability) was 81% higher (OR = 1.81; 95% CI, 1.72–1.91) for nephropathy, 17% higher (OR = 1.17; 95% CI, 1.13–1.21) for retinopathy, 30% higher (OR = 1.30; 95% CI, 1.25–1.35) for neuropathy, and 19% higher (OR = 1.19; 95% CI, 1.15–1.23) for any complication. When used as a continuous variable, SBP variability was linearly associated with increased risks of all four outcomes (see Table B).
Table 3.
Odds ratios and 95% confidence intervals for incident microvascular complications in SBP VIM Quartiles 2–4 compared to Quartile 1, all person-years at risk in 2004–2008.a
| Complication | N | VIM quartiles compared to Quartile 1 (Least variable)
|
P for Trend | |||||
|---|---|---|---|---|---|---|---|---|
| 2 | 3 | 4 (Most variable) | ||||||
| Nephropathy | 636,129 | 1.244 | (1.177–1.316) | 1.435 | (1.359–1.515) | 1.813 | (1.720–1.910) | <0.001 |
| Retinopathy | 460,006 | 1.056 | (1.018–1.095) | 1.114 | (1.075–1.155) | 1.171 | (1.129–1.214) | <0.001 |
| Neuropathy | 476,462 | 1.169 | (1.128–1.211) | 1.270 | (1.226–1.315) | 1.299 | (1.253–1.346) | <0.001 |
| Any complication | 322,837 | 1.119 | (1.084–1.156) | 1.167 | (1.129–1.206) | 1.187 | (1.147–1.227) | <0.001 |
SBP VIM = systolic blood pressure variability independent of the mean; models were adjusted for age in 2003, male, race/ethnicity, marital status, body mass index, mean A1c, diabetes duration >5 years in 2003, diabetes treatment (none, oral medications alone, insulin alone or oral and insulin medications), statin use, antihypertensive use, and mean systolic blood pressure.
In sensitivity analyses, we found strong graded relationships between SBP variability and all outcomes for both persons with elevated mean BP (≥130/80 mm Hg) and those with normal mean BP (P for trend <0.05 for all models; Table 4). A subgroup analysis of patients with BP <130/80 mg Hg stratified by antihypertensive medication use (Table 4) showed a significant graded relationship between SBP variability and the incidence of nephropathy and neuropathy for both antihypertensive users and non-users (P for trend ≤0.05 for all outcomes). For retinopathy and any complication, there was a significant graded relationship (P for trend <0.001) among users but not among non-users. In another sensitivity analysis stratified by the median number of BP measures (six measures per year), we found that both groups (≤6 measures and 7 or more) showed significant graded relationships between SBP variability and all outcomes with P for trend <0.001 (Table 4).
Table 4.
Odds ratios and 95% confidence intervals for incident microvascular complications in SBP VIM Quartiles 2–4 compared to Quartile 1, stratified by BP level, medication use, and number of BP measures.a
| N | VIM quartiles compared to Quartile 1 (Least variable)
|
P for Trend | ||||||
|---|---|---|---|---|---|---|---|---|
| 2 | 3 | 4 (Most variable) | ||||||
| Nephropathy | ||||||||
| By BP level | ||||||||
| < 130/80 mm Hg | 220,180 | 1.275 | (1.144–1.421) | 1.499 | (1.348–1.667) | 1.876 | (1.686–2.088) | <0.001 |
| ≥ 130/80 mm Hg | 415,949 | 1.248 | (1.168–1.333) | 1.437 | (1.348–1.532) | 1.824 | (1.716–1.938) | <0.001 |
| By medication use | ||||||||
| No | 37,435 | 1.243 | (0.906–1.707) | 1.214 | (0.862–1.712) | 1.783 | (1.240–2.564) | 0.005 |
| Yes | 182,745 | 1.266 | (1.126–1.423) | 1.495 | (1.334–1.676) | 1.826 | (1.629–2.046) | <0.001 |
| By number of measures | ||||||||
| 6 or fewer | 350,498 | 1.067 | (0.987–1.154) | 1.164 | (1.076–1.259) | 1.444 | (1.343–1.553) | <0.001 |
| 7 or more | 285,631 | 1.305 | (1.194–1.426) | 1.512 | (1.387–1.647) | 2.015 | (1.852–2.192) | <0.001 |
| Retinopathy | ||||||||
| By BP level | ||||||||
| <130/80 mm Hg | 162,302 | 1.061 | (0.994–1.131) | 1.120 | (1.049–1.195) | 1.155 | (1.078–1.238) | <0.001 |
| ≥130/80 mm Hg | 297,704 | 1.052 | (1.006–1.099) | 1.107 | (1.060–1.157) | 1.163 | (1.114–1.245) | <0.001 |
| By medication use | ||||||||
| No | 29,156 | 1.130 | (0.981–1.303) | 1.136 | (0.972–1.327) | 1.103 | (0.918–1.326) | 0.159 |
| Yes | 133,146 | 1.044 | (0.971–1.123) | 1.116 | (1.035–1.196) | 1.154 | (1.070–1.245) | <0.001 |
| By number of measures | ||||||||
| 6 or fewer | 259,196 | 1.042 | (0.994–1.092) | 1.039 | (0.989–1.091) | 1.091 | (1.040–1.145) | 0.001 |
| 7 or more | 200,810 | 1.057 | (0.994–1.124) | 1.161 | (1.093–1.233) | 1.239 | (1.165–1.319) | <0.001 |
| Neuropathy | ||||||||
| By BP level | ||||||||
| <130/80 mm Hg | 159,767 | 1.222 | (1.151–1.298) | 1.270 | (1.194–1.350) | 1.359 | (1.275–1.298) | <0.001 |
| ≥130/80 mm Hg | 316,695 | 1.145 | (1.096–1.196) | 1.277 | (1.226–1.333) | 1.281 | (1.226–1.338) | <0.001 |
| By medication use | ||||||||
| No | 29,252 | 1.303 | (1.134–1.498) | 1.324 | (1.138–1.541) | 1.180 | (0.983–1.417) | 0.006 |
| Yes | 130,515 | 1.199 | (1.121–1.283) | 1.252 | (1.170–1.339) | 1.364 | (1.273–1.463) | <0.001 |
| By number of measures | ||||||||
| 6 or fewer | 278,827 | 1.071 | (1.018–1.126) | 1.146 | (1.089–1.207) | 1.157 | (1.099–1.218) | <0.001 |
| 7 or more | 197,635 | 1.082 | (1.023–1.145) | 1.159 | (1.097–1.225) | 1.239 | (1.171–1.311) | <0.001 |
| Any complication | ||||||||
| By BP Level | ||||||||
| <130/80 mm Hg | 111,097 | 1.136 | (1.074–1.201) | 1.182 | (1.116–1.252) | 1.179 | (1.108–1.254) | <0.001 |
| ≥130/80 mm Hg | 211,740 | 1.111 | (1.068–1.156) | 1.160 | (1.114–1.207) | 1.185 | (1.138–1.233) | <0.001 |
| By medication use | ||||||||
| No | 22,529 | 1.253 | (1.109–1.416) | 1.187 | (1.036–1.360) | 1.096 | (0.928–1.293) | 0.080 |
| Yes | 88,568 | 1.105 | (1.037–1.177) | 1.172 | (1.099–1.249) | 1.175 | (1.098–1.258) | <0.001 |
| By number of measures | ||||||||
| 6 or fewer | 195,667 | 1.049 | (1.006–1.095) | 1.069 | (1.022–1.117) | 1.078 | (1.031–1.127) | <0.001 |
| 7 or more | 127,170 | 1.109 | (1.050–1.172) | 1.161 | (1.100–1.226) | 1.233 | (1.165–1.306) | <0.001 |
SBP VIM = systolic blood pressure variability independent of the mean; all models were adjusted for demographic factors (age in 2003, male, race/ethnicity, marital status), clinical factors (body mass index, mean A1c, diabetes duration >5 years in 2003, diabetes treatment, statin use), and mean SBP. Each outcome was stratified by blood pressure level (< or ≥130/80 mg Hg), antihypertensive medication use (yes or no for persons with BP <130/80 mg Hg only), and number of BP measures (≤ 6 or >6).
The method identifying complications produced considerably different incidence rates but the graded relationship between SBP variability and outcomes persisted in all models, even though the magnitude of association varied from method to method (Table C).
In supplemental analyses, we found that SBP variability was also significantly associated with higher risks of mortality and macrovascular disease (Table D). Compared to those in Quartile 1, patients in Quartile 4 had 2.3 times higher mortality (OR = 2.32; 95% CI, 2.22–2.42) and 1.6 times higher risk of any macrovascular disease (OR = 1.61; 95% CI, 1.55–1.67).
4. Discussion
Our results show a significant graded relationship between SBP variability and the incidence of microvascular complications among non-elderly persons with diabetes. The relationship was the strongest for nephropathy. We also found that patients with the highest variability (Quartile 4) were 17% and 30% more likely to experience incident retinopathy and neuropathy, respectively, than those with the lowest variability (Quartile 1). The risk of any microvascular complication, a combined endpoint, was about 19% higher in the highest variability group than the lowest. The gradient relationship persisted in all sensitivity analyses stratified by mean BP (≥130/80 mm Hg or below) or by the total number of BP measures taken during a given year (≤6 and 7 or more). Three different methods of identifying the endpoints did not substantially change the results either, adding confidence to the robustness of our results. Our analysis further revealed that SBP variability increased the risk of mortality and macrovascular disease (Table D), consistent with previous studies (Hata, Arima, Rothwell, et al., 2013; Rothwell et al., 2010a; Stevens, Wood, Koshiaris, et al., 2016).
These relationships held not only in patients who had BP ≥130/80 mm Hg but also in patients whose BP was controlled <130/80 mm Hg with antihypertensive medications (American Diabetes Association, 2014). Furthermore, patients who had BP <130/80 mm Hg and did not take any antihypertensive medications also demonstrated a similar graded relationship between SBP variability and both nephropathy and neuropathy. These are normotensive individuals who clinicians are not likely to treat for blood pressure but our results suggest that they may also be vulnerable to the harmful effects of SBP variability.
Since 2010, several studies have reported a significant relationship between SBP variability and nephropathy (Hata et al., 2013; Kilpatrick et al., 2010; Noshad et al., 2014; Okada, Fukui, et al., 2012; Okada et al., 2013; Takao et al., 2014). This is the first study that demonstrated significant graded relationships between SBP variability and diabetic retinopathy and neuropathy. Retinopathy has been examined in three previous studies. The first was a post-hoc analysis of the Diabetes Control and Complication Trial data for patients with type 1 diabetes (Kilpatrick et al., 2010). The second was a case series of about 600 patients with type 2 diabetes and without diabetic retinopathy at baseline (Takao et al., 2014). Neither of these studies found any relationship between the two. A third study by Hata et al. showed a positive relationship but did not achieve a statistical significance for a linear trend (Hata et al., 2013). In contrast, our study showed a significant and robust graded relationship between the SBP variability and retinopathy in our main analysis and all stratified analyses except for the individuals who had BP controlled at <130/80 mm Hg without using antihypertensive medications.
Our results regarding neuropathy are novel and clinically significant given its high prevalence among diabetic patients. We could not find any previous study that examined its relationship with SBP variability. This may be because the general relationship between high BP and diabetic neuropathy is not well established yet (Cho, Mold, & Roberts, 2006; Forrest, Maser, Pambianco, Becker, & Orchard, 1997; Stella, Ellis, Maser, & Orchard, 2000; UK Prospective Diabetes Study Group, 1998). Our results suggest an insignificant association between SBP mean and the risk of neuropathy (see Table A).
There have been several explanations of the relationship between excessive SBP variability and cardiovascular outcomes, including arterial stiffness, abnormal autonomic function, and direct damage to end organs (Hata et al., 2013; Parati, Liu, Ochoa, & Bilo, 2013; Rothwell et al., 2010a; Su, 2006). But poor adherence to antihypertensive therapies does not appear to be the main explanation for excessive SBP variability among hypertensive patients (Krousel-Wood et al., 2009). Our results on neuropathy and patients with normotensive BP is an important new clinical observation that may, with additional research, add to the current understanding of the pathophysiological mechanisms by which SBP variability can do direct damages to the nervous system.
Our results have several implications for clinical practice. In addition to the traditional atherosclerotic cardiovascular diseases, microvascular conditions such as nephropathy, retinopathy, and neuropathy may need to be closely monitored for patients with high SBP variability. Treatment for excessive SBP variability in addition to high mean SBP may be necessary for prevention of these complications for hypertensive patients. But, even among individuals whose BP are controlled at <130/80 mm Hg with antihypertensive medications, we observed a significantly elevated risks of all three complications. This means that treatment for hypertension may need to be augmented with a therapy for reducing SBP variability (Rothwell et al., 2010b; Su, 2006). Furthermore, our finding that diabetic patients who have normal BP and are not treated with antihypertensive medications are also subject to the harmful effects of high SBP variability on kidney and peripheral nervous systems suggests that excessive SBP variability alone can produce end organ damages (Su, 2006; Miao, Xie, Zhan, & Su, 2006). Even though they may not be treated for high mean BP, patients may need to be treated for excessive SBP variability (Rothwell et al., 2010b; Su, 2006). Lastly, we observed significantly elevated risks of all three complications between Quartiles 1 and 2 which were separated by a mean within-subject SBP standard deviation of about 4 mm Hg (Table 1). This suggests that clinicians may need to begin therapy for SBP variability as small as 4 mm Hg in standard deviation.
Before any changes in clinical practice are attempted, however, more research is needed to rigorously examine causality and to better understand the underlying pathophysiological mechanisms. Future interventional trials are also needed to examine which antihypertensive medications are the most efficacious in reducing SBP variability or whether lifestyle modification can reduce SBP variability by itself or in addition to medications. Future research is further needed to examine whether there are other conditions such as foot ulcers and amputations, autonomic neuropathy, peripheral vascular disease, or other central nervous system diseases (Epstein et al., 2013) that may be associated with BP variability.
This study has several limitations. This is an observational study in which unmeasured confounding can be a source of bias. Our data need to be carefully interpreted. Second, disease coding in ICD-9-CM may not be accurate and this might have affected our ability to identify diabetic microvascular complications. Previous studies showed 87%–97% sensitivity for retinopathy compared to medical records (Muir, Gupta, Gill, & Stein, 2013) and 93%–99% specificity for nephropathy in the VA healthcare system (Kern, Maney, Miller, et al., 2006). Identification of diabetic retinopathy based on a single code and the other two methods we considered in this study differed considerably. Many patients diagnosed with diabetic retinopathy had no follow-up visits in the year following their diagnosis, likely due to slow disease progression and/or absence of symptoms. In our review of electronic medical records for 15 randomly selected patients with one code for incident diabetic retinopathy in 2008, we found that 11 (73%) had positive diagnoses on clinical examination. This shows that the majority of the patients identified by the single code method indeed had the condition. Our sensitivity analyses further indicate that our results are robust to the method of identification. Third, the blood pressure data used in this study were collected by VA hospitals and outpatient clinics across the country and there is currently no information on how standardized the measurements were. However, the VA follows a national guideline for measuring blood pressure consistent with the JNC7 guideline (Chobanian, Bakris, Black, et al., 2003; The Diagnosis and Management of Hypertension Working Group, 2014) and a uniform policy in maintaining and calibrating BP devices according to the manufacturer recommendations. A validation study found that the BP measures in the VA were reliable to assess quality of hypertension care (Borzecki, Wong, Hickey, Ash, & Berlowitz, 2004). Fourth, we did not have access to lifestyle factors such as smoking, diet, and physical activity that may have affected diabetes management in general. Finally, our cohort was limited to the non-elderly veterans with diabetes who were overwhelmingly male and so the generalizability of our results is limited.
In conclusion, our results show a significant gradient relationship between SBP variability and incidence of microvascular complications for non-elderly patients with diabetes.
Supplementary Material
Acknowledgments
The authors gratefully acknowledge the financial support from the US Department of Veterans Affairs, Health Services Research and Development Services (LIP 42-151), and the Agency for Healthcare Research and Quality (R01HS018542). The paper presents the findings and conclusions of the authors; it does not necessarily represent the Department of Veterans Affairs, Health Services Research and Development Service, or the Agency for Healthcare Research and Quality.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jdiacomp.2016.09.003.
Footnotes
Conflict of Interest: There is no conflict of interest to declare.
Contributor Statements
M.S., N.E. and E.B. conceived the design of the study, M.S. researched data and wrote manuscript, and Z.H. provided statistical support. All authors contributed to the interpretation of the data, critically reviewed the manuscript, and approved the final version of the manuscript. M.S. is the guarantor of this study, had full access to all the data used in this study, and takes responsibility for the integrity of the data and the accuracy of the data analysis.
References
- American Diabetes Association. Diabetes Care. 2014;37(Suppl. 1):S14–S80. doi: 10.2337/dc14-S014. [DOI] [PubMed] [Google Scholar]
- American Diabetes Association. Standards of Medical Care in Diabetes – 2015. Diabetes Care. 2015;38(Supplement 1):S1–S87. [Google Scholar]
- Borzecki AM, Wong AT, Hickey EC, Ash AS, Berlowitz DR. Can we use automated data to assess quality of hypertension care? The American Journal of Managed Care. 2004;10(7 Pt 2):473–479. [PubMed] [Google Scholar]
- Cho DY, Mold JW, Roberts M. Further investigation of the negative association between hypertension and peripheral neuropathy in the elderly: an Oklahoma Physicians Resource/Research Network (OKPRN) Study. Journal of the American Board of Family Medicine. 2006;19(3):240–250. doi: 10.3122/jabfm.19.3.240. [DOI] [PubMed] [Google Scholar]
- Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL, Jr, Roccella EJ. Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Hypertension. 2003;42(6):1206–1252. doi: 10.1161/01.HYP.0000107251.49515.c2. [DOI] [PubMed] [Google Scholar]
- Curb JD, Pressel SL, Cutler JA, Savage PJ, Applegate WB, Black H, Stamler J. Effect of diuretic-based antihypertensive treatment on cardiovascular disease risk in older diabetic patients with isolated systolic hypertension. Systolic Hypertension in the Elderly Program Cooperative Research Group. Journal of the American Medical Association. 1996;276(23):1886–1892. [PubMed] [Google Scholar]
- Epstein NU, Lane KA, Farlow MR, Risacher SL, Saykin AJ, Gao S, Alzheimer’s Cognitive dysfunction and greater visit-to-visit systolic blood pressure variability. Journal of the American Geriatrics Society. 2013;61(12):2168–2173. doi: 10.1111/jgs.12542. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Forrest KY, Maser RE, Pambianco G, Becker DJ, Orchard TJ. Hypertension as a risk factor for diabetic neuropathy: a prospective study. Diabetes. 1997;46(4):665–670. doi: 10.2337/diab.46.4.665. [DOI] [PubMed] [Google Scholar]
- Hata J, Arima H, Rothwell PM, Woodward M, Zoungas S, Anderson C, Group, A. C. Effects of visit-to-visit variability in systolic blood pressure on macrovascular and microvascular complications in patients with type 2 diabetes mellitus: the ADVANCE trial. Circulation. 2013;128(12):1325–1334. doi: 10.1161/CIRCULATIONAHA.113.002717. [DOI] [PubMed] [Google Scholar]
- Kern EF, Maney M, Miller DR, Tseng CL, Tiwari A, Rajan M, Pogach L. Failure of ICD-9-CM codes to identify patients with comorbid chronic kidney disease in diabetes. Health Services Research. 2006;41(2):564–580. doi: 10.1111/j.1475-6773.2005.00482.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kilpatrick ES, Rigby AS, Atkin SL. The role of blood pressure variability in the development of nephropathy in type 1 diabetes. Diabetes Care. 2010;33(11):2442–2447. doi: 10.2337/dc10-1000. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Krousel-Wood M, Islam T, Webber LS, Re RN, Morisky DE, Muntner P. New medication adherence scale versus pharmacy fill rates in seniors with hypertension. The American Journal of Managed Care. 2009;15(1):59–66. [PMC free article] [PubMed] [Google Scholar]
- Lee TA, Shields AE, Vogeli C, Gibson TB, Sohn MW, Marder WD, Weiss K. Mortality rate in veterans with multiple chronic conditions. Journal of General Internal Medicine. 2007;22(Suppl. 3):403–407. doi: 10.1007/s11606-007-0277-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Miao CY, Xie HH, Zhan LS, Su DF. Blood pressure variability is more important than blood pressure level in determination of end-organ damage in rats. Journal of Hypertension. 2006;24(6):1125–1135. doi: 10.1097/01.hjh.0000226203.57818.88. [DOI] [PubMed] [Google Scholar]
- Miller DR, Safford MM, Pogach LM. Who has diabetes? Best estimates of diabetes prevalence in the Department of Veterans Affairs based on computerized patient data. Diabetes Care. 2004;27(Suppl. 2):B10–B21. doi: 10.2337/diacare.27.suppl_2.b10. [DOI] [PubMed] [Google Scholar]
- Muir KW, Gupta C, Gill P, Stein JD. Accuracy of international classification of diseases, ninth revision, clinical modification billing codes for common ophthalmic conditions. JAMA Ophthalmology. 2013;131(1):119–120. doi: 10.1001/jamaophthalmol.2013.577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Muntner P, Shimbo D, Tonelli M, Reynolds K, Arnett DK, Oparil S. The relationship between visit-to-visit variability in systolic blood pressure and all-cause mortality in the general population: findings from NHANES III, 1988 to 1994. Hypertension. 2011;57(2):160–166. doi: 10.1161/HYPERTENSIONAHA.110.162255. [DOI] [PubMed] [Google Scholar]
- Muntner P, Whittle J, Lynch AI, Colantonio LD, Simpson LM, Einhorn PT, Oparil S. Visit-to-visit variability of blood pressure and coronary heart disease, stroke, heart failure, and mortality: a cohort study. Annals of Internal Medicine. 2015;163(5):329–338. doi: 10.7326/M14-2803. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Noshad S, Mousavizadeh M, Mozafari M, Nakhjavani M, Esteghamati A. Visit-to-visit blood pressure variability is related to albuminuria variability and progression in patients with type 2 diabetes. Journal of Human Hypertension. 2014;28(1):37–43. doi: 10.1038/jhh.2013.36. [DOI] [PubMed] [Google Scholar]
- Okada H, Fukui M, Tanaka M, Inada S, Mineoka Y, Nakanishi N, Nakamura N. Visit-to-visit variability in systolic blood pressure is correlated with diabetic nephropathy and atherosclerosis in patients with type 2 diabetes. Atherosclerosis. 2012a;220(1):155–159. doi: 10.1016/j.atherosclerosis.2011.10.033. [DOI] [PubMed] [Google Scholar]
- Okada H, Fukui M, Tanaka M, Matsumoto S, Mineoka Y, Nakanishi N, Nakamura N. Visit-to-visit blood pressure variability is a novel risk factor for the development and progression of diabetic nephropathy in patients with type 2 diabetes. Diabetes Care. 2013;36(7):1908–1912. doi: 10.2337/dc12-2087. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Okada T, Matsumoto H, Nagaoka Y, Nakao T. Association of home blood pressure variability with progression of chronic kidney disease. Blood Pressure Monitoring. 2012b;17(1):1–7. doi: 10.1097/MBP.0b013e32834f7125. [DOI] [PubMed] [Google Scholar]
- Parati G, Liu X, Ochoa JE, Bilo G. Prognostic relevance of blood pressure variability: role of long-term and very long-term blood pressure changes. Hypertension. 2013;62(4):682–684. doi: 10.1161/HYPERTENSIONAHA.113.01801. [DOI] [PubMed] [Google Scholar]
- Rothwell PM, Howard SC, Dolan E, O’Brien E, Dobson JE, Dahlof B, Poulter NR. Prognostic significance of visit-to-visit variability, maximum systolic blood pressure, and episodic hypertension. Lancet. 2010a;375(9718):895–905. doi: 10.1016/S0140-6736(10)60308-X. [DOI] [PubMed] [Google Scholar]
- Rothwell PM, Howard SC, Dolan E, O’Brien E, Dobson JE, Dahlof B, Investigators, M. R. C. T. Effects of beta blockers and calcium-channel blockers on within-individual variability in blood pressure and risk of stroke. Lancet Neurology. 2010b;9(5):469–480. doi: 10.1016/S1474-4422(10)70066-1. [DOI] [PubMed] [Google Scholar]
- Schutte R, Thijs L, Liu YP, Asayama K, Jin Y, Odili A, Staessen JA. Within-subject blood pressure level– not variability– predicts fatal and nonfatal outcomes in a general population. Hypertension. 2012;60(5):1138–1147. doi: 10.1161/HYPERTENSIONAHA.112.202143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sohn MW, Budiman-Mak E, Stuck RM, Siddiqui F, Lee TA. Diagnostic accuracy of existing methods for identifying diabetic foot ulcers from inpatient and outpatient datasets. Journal of Foot and Ankle Research. 2010;3:27. doi: 10.1186/1757-1146-3-27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Stella P, Ellis D, Maser RE, Orchard TJ. Cardiovascular autonomic neuropathy (expiration and inspiration ratio) in type 1 diabetes. Incidence and predictors. Journal of Diabetes and its Complications. 2000;14(1):1–6. doi: 10.1016/s1056-8727(00)00054-4. [DOI] [PubMed] [Google Scholar]
- Stevens SL, Wood S, Koshiaris C, Law K, Glasziou P, Stevens RJ, McManus RJ. Blood pressure variability and cardiovascular disease: systematic review and meta-analysis. BMJ. 2016;354:i4098. doi: 10.1136/bmj.i4098. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Su DF. Treatment of hypertension based on measurement of blood pressure variability: lessons from animal studies. Current Opinion in Cardiology. 2006;21(5):486–491. doi: 10.1097/01.hco.0000240587.14463.58. [DOI] [PubMed] [Google Scholar]
- Takao T, Matsuyama Y, Yanagisawa H, Kikuchi M, Kawazu S. Visit-to-visit variability in systolic blood pressure predicts development and progression of diabetic nephropathy, but not retinopathy, in patients with type 2 diabetes. Journal of Diabetes and its Complications. 2014;28(2):185–190. doi: 10.1016/j.jdiacomp.2013.11.003. [DOI] [PubMed] [Google Scholar]
- Defense TDoVAatDo, editor. The Diagnosis and Management of Hypertension Working Group. VA/DoD clinical practice guideline for the diagnosis and management of hypertension in the primary care setting. 2014. [Google Scholar]
- Tuomilehto J, Rastenyte D, Birkenhager WH, Thijs L, Antikainen R, Bulpitt CJ, Fagard R. Effects of calcium-channel blockade in older patients with diabetes and systolic hypertension. Systolic Hypertension in Europe Trial Investigators. The New England Journal of Medicine. 1999;340(9):677–684. doi: 10.1056/NEJM199903043400902. [DOI] [PubMed] [Google Scholar]
- UK Prospective Diabetes Study Group. Tight blood pressure control and risk of macrovascular and microvascular complications in type 2 diabetes: UKPDS 38. BMJ. 1998;317(7160):703–713. [PMC free article] [PubMed] [Google Scholar]
- Ushigome E, Fukui M, Hamaguchi M, Senmaru T, Sakabe K, Tanaka M, Nakamura N. The coefficient variation of home blood pressure is a novel factor associated with macroalbuminuria in type 2 diabetes mellitus. Hypertension Research: Official Journal of the Japanese Society of Hypertension. 2011;34(12):1271–1275. doi: 10.1038/hr.2011.128. [DOI] [PubMed] [Google Scholar]
- Veterans Affairs Information Resource Center. VIReC research user guide: FY2004 VHA medical SAS inpatient datasets 2005a [Google Scholar]
- Veterans Affairs Information Resource Center. VIReC research user guide: FY2004 VHA medical SAS outpatient datasets 2005b [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.