Postural changes in blood pressure among patients with diabetes attending a referral hospital in southwestern Uganda: a cross-sectional study

This was a cross-sectional study, conducted at the Diabetes and Endocrinology Clinic of Mbarara Regional Referral Hospital (MRRH) from November 2018 to April 2019. The detailed methods for the parent study that assessed the overall burden of cardiovascular autonomic neuropathy among individuals with diabetes in ambulatory care, and a detailed description of the study population at inclusion, have been published previously [22]. In brief, we included individuals with diabetes aged 18–65 years. We excluded patients with known underlying hepatic, renal, or cardiac diseases. We also excluded patients with known ECG or electrolyte abnormalities, those with acute febrile illnesses, and those actively taking antihypertensive medications such as calcium channel blockers, beta blockers and diuretics in the previous 24 h.

We collected clinical and demographic data, using an interviewer administered structured questionnaire. Waist circumference was measured at the level of umbilicus with inelastic tapeline (to the nearest 0.1 cm), at the end of normal expiration. Height and weight of study participants were measured to the nearest 0.1 cm and 0.1 kg respectively. We calculated body mass index as the weight (kg) divided by the square of height (m²). Measurements for glycosylated hemoglobin (HbA1c) were performed using an automated high performance liquid chromatography analyzer (Cobas Integra 400, Roche diagnostics, Basel, Switzerland), at Lancet laboratories. Measurements of fasting blood glucose were obtained under aseptic technique, using a Freestyle Glucometer (Abbott Diabetes Care Inc., Maidenhead, UK) after at least 8 h of fasting, by means of capillary blood obtained by finger prick. Diabetes was defined in participants with fasting capillary glucose of ≥ 7.0 mmol/L or those who were already on medication for diabetes mellitus, as per recognized criteria [23]. Diabetic retinopathy was diagnosed using direct ophthalmoscopy and classified as non-proliferative or proliferative diabetic retinopathy, by an ophthalmologist.

We performed resting 12 lead ECG recordings using a portable ECG machine (Edan Instruments, Inc., Hessen, Germany) on study participants, and measured the QT interval from the beginning of the earliest onset of the QRS complex to the end of the T wave, where it crosses the isoelectric line. We used the Bazette’s formula to calculate a heart rate–corrected QT (QTc) [24]. We consider a cut-off value of the QTc interval > 440 ms to be elevated [25].

All BP recordings were done using an automatic sphygmomanometer in the upper arm (Omron HEM 705 LP, Omron Healthcare, Inc., Bannockburn, IL, USA). The protocol for assessment of OH and OHT included supine BP and standing BP measurements. Supine BP was recorded after 5 min of rest. Standing BP was measured within 3 min of assuming upright standing position. We defined OH in participants with either a ≥ 20 mmHg drop in systolic BP (SBP) or ≥ 10 mmHg drop in diastolic BP (DBP) when assuming an upright position [26]. To define OHT, we considered the cut-off for abnormal systolic orthostatic blood pressure responses of ≥ 20 mmHg, which has been previously proposed for hypertensive orthostatic BP responses [1]. We thus defined systolic OHT in participants with a rise in SBP ≥ 20 mmHg, with or without an accompanying DBP rise; systolic OH was defined in participants with a drop of SBP ≥ 20 mmHg, with or without an accompanying DBP drop. To define diastolic orthostatic hypotensive responses, we considered the cut-off was ≥ 10 mmHg [1] and applied this cut-off for diastolic OHT, since there is no consensus for the numerical threshold for this entity [18]. Thus, we defined diastolic OHT in participants with an increase in diastolic BP ≥ 10 mmHg, with or without an accompanying SBP increase, and diastolic OH in participants with a drop in diastolic BP ≥ 10 mmHg, with or without an accompanying drop in SBP Accordingly, a normal diastolic orthostatic BP response was defined in participants with a change of DBP within the range of − 9 to + 9 mmHg, whereas a normal systolic orthostatic BP response was defined in participants with a change of SBP within the range of − 19 to + 19 mmHg.

Conventional BP measurements were done with participants in sitting positions. We defined hypertension in participants whose BP values were ≥ 140/90 mmHg or who were already on medications for high BP [27]. We calculated pulse pressure as the difference between systolic and diastolic BP measurements. Although there is no widely recognized cut-off for defining high pulse pressure, we adopted the previously suggested threshold of ≥ 65 mm Hg [28].

For the parent study, which was designed to estimate the prevalence of autonomic neuropathy, we estimated a sample size of 296 participants around a prevalence estimate of cardiovascular autonomic neuropathy (CAN) of 20% [29], with a 5% precision at a 95% level of confidence, after inflation for 10% non-response rate, using Epi Info (version 7.1.4.0, CDC, Atlanta, US). Study data were entered data into EpiData3. (EpiData, Odense, Denmark). We used Stata, version 13 (StataCorp, College Station, Texas, USA) for all statistical analyses.

Our primary outcomes were OH and OHT. First, we determined the prevalence of OH and OHT, as the proportion of participants that met their definitions. Next, we compared differences in clinical and demographic characteristics between participants with OH and OHT and those without. Continuous normally distributed data (presented as mean ± standard deviation) were compared using a two-tailed independent t-test; while continuous non-normally distributed data (presented as median ± inter-quartile range) were compared using Wilcoxon rank-sum test. We compared categorical variables using the chi-squared test. Finally, we performed univariate and multivariate logistic regression analyses to identify factors associated with OH and OHT. All variables with a P-value less than 0.1, were included in the multivariate models and adjusted for sex, age, BMI, HbA1c, QTc interval, and duration of diabetes, based on biological plausibility. Pulse pressure—derived from SBP and DPB— and history of hypertension, were eliminated from the final multivariate model because of collinearity. We considered statistical significance at a threshold of P < 0.05.

The clinical and demographic characteristics of the total cohort have been described previously [22]. In brief, we enrolled a total of 299 participants, with a mean age of 50 years (SD ± 9.8), mean duration of diabetes of 5.8 (SD ± 5.9) years, and mean HbA1c of 9.7% (SD ± 2.6); 70% were female.

Overall, abnormal postural changes in BP were detected in 66 participants (22.1%). Of the 299 participants assessed orthostatic BP responses, 52 (17.4%; 95% CI 13.3–22.2%) met the definition of OH; 43 (14.4%; 95% CI 10.6–18.9%) were classified as having OHT (Table 1).

Diastolic hypertension was found in 38 participants (12.7%), systolic hypotension in 23 participants (7.7%), while diastolic hypotension and systolic hypertension were found in 10 participants (3.3%) each. The differences between blood pressures from supine to standing followed a normal distribution pattern for both systolic and diastolic BP measurements are shown in Fig. 1. The mean (SD) change for the systolic BP was 3 (± 13) mmHg; for the DBP, the mean change (SD) was -4 (± 8) mmHg.

We found crude differences in the mean age (P = 0.002), history of hypertension (P = 0.002), history of palpitations (P = 0.002), diabetic retinopathy (P < 0.001), mean QTc interval (P = 0.006) and median duration of diabetes (P = 0.031) between those with and without OH. Participants with OH also had significantly higher mean resting systolic BP (P < 0.001), higher mean resting diastolic BP (P = 0.004) and higher mean pulse pressure (P < 0.001) compared with their counterparts with no OH.

The proportion of female participants (P = 0.011) and those who reported history of palpitations (P < 0.001) was significantly higher in the OHT group than in the group without OHT, as shown in Table 2.

Participants with OHT had significantly higher mean BMI (P = 0.015), higher mean waist circumference (P = 0.019), higher mean pulse pressure (P < 0.001), higher mean resting systolic BP (P = 0.001), and higher mean QTc (P = 0.049) than their counterparts (Table 2).

In multivariate analysis (Table 3), the factors associated with diagnosis of diabetic OH were older age of 51–65 years (OR = 2.40; 95% CI 1.02–5.67, P = 0.046), presence of diabetic retinopathy (OR = 2.51; 95% CI 1.14–5.53, P = 0.022), history of palpitations (OR = 2.31; 95% CI 1.08–4.92, P = 0.031), and higher resting systolic BP of ≥ 140 mmHg (OR = 3.14; 95% CI 1.31–8.7.56, P = 0.011).

Having history of palpitations (OR = 3.14; 95% CI 1.42–6.95, P = 0.005) and higher resting SBP ≥ 140 mmHg (OR = 2.01; 95% CI 1.10–4.42, P = 0.043) were associated with presence of OHT in the multivariate analysis (Table 3).

The findings from this study should be interpreted in consideration of our study limitations. First, our study population lacked individuals older than 65 years, based on the exclusion criteria of the parent study which assessed the overall burden of cardiovascular autonomic neuropathy among individuals with diabetes in ambulatory care. This may have led to underestimation of the prevalence of the OH and OHT, and biased our associations towards the null. Generalizability should be restricted similarly to those with diabetes and in ambulatory care in similar peri-urban African settings. Second, the cross-sectional study design limited us from assigning time causal directionally between our exposures and outcomes. Third, we did not comprehensively assess complications of OH and OHT including orthostatic intolerance and associated falls. Thus, we cannot evaluate the clinical implications of these abnormal orthostatic BP responses in the study population. Further studies are required to assess the prognostic and clinical implications of these orthostatic abnormalities in this patient population in the region. Despite these limitations, our study is among the first studies to provide useful epidemiological data on OH and OHT among diabetic individuals in sub-Saharan African region.