Subclinical left ventricular systolic dysfunction on speckle-tracking echocardiography in hypertensive outpatients with preserved ejection fraction: prevalence, phenotypes and correlates in Cameroon.
Dysfonction systolique ventriculaire gauche infraclinique en speckle-tracking chez des patients hypertendus ambulatoires avec fraction d’éjection ventriculaire gauche préservée : prévalence, phénotypes et facteurs associés au Cameroun.
H BA¹²*, MD SALI BOUHARI², S DJALLO², TIMNOU BEKOUTI J², NGANOU-GNINDJIO CN²,³.
RESUME
Introduction : La fraction d’éjection du ventricule gauche (FEVG) reste dans les limites de la normale longtemps après le début de l’atteinte myocardique hypertensive. Le strain longitudinal global (SLG) détecte cette atteinte plus tôt, car les fibres longitudinales sous-endocardiques sont les premières touchées. Les données de strain restent limitées en Afrique subsaharienne, avec des travaux récents au Cameroun et au Togo, et aucune n’a été rapportée dans le Nord du Cameroun. Cette étude visait à déterminer la prévalence, les phénotypes et les facteurs associés de la dysfonction systolique ventriculaire gauche infraclinique chez des patients hypertendus suivis en consultation externe dans le Nord du Cameroun.
Méthodes : Nous avons mené une étude transversale analytique dans le service de cardiologie de l’Hôpital général de Garoua de mai à septembre 2026. Ont été inclus de façon consécutive les adultes âgés d’au moins 21 ans, traités pour une hypertension artérielle essentielle, ayant une FEVG ≥ 50 % et sans symptôme ni signe d’insuffisance cardiaque. Le SLG a été mesuré par speckle tracking bidimensionnel sur 17 segments et exprimé en valeur absolue ; une altération était définie par un |SLG| inférieur à 18 %. La reproductibilité intra- et inter-observateur a été évaluée chez 30 patients tirés au sort, au moyen de coefficients de corrélation intraclasse, d’une analyse de Bland–Altman, de coefficients de variation et du kappa de Cohen pour le classement altéré/normal. Les rapports de prévalence ont été estimés par régression de Poisson pénalisée de Firth à variance robuste, avec correction pour petits échantillons de Morel–Bokossa–Neerchal, en ajustant sur des covariables choisies a priori sur des bases cliniques. Les phénotypes cliniques ont été identifiés par analyse des correspondances multiples suivie d’une classification ascendante hiérarchique selon la méthode de Ward. La relation entre la FEVG et le SLG a été modélisée par régression linéaire avec des erreurs-types robustes à l’hétéroscédasticité (HC3).
Résultats : Sur 119 patients évalués, 105 ont été analysés. L’âge moyen était de 56,4 ans ± 10,5 et 58 participants (55,2 %) étaient des femmes. La durée médiane de l’hypertension était de 3 ans (EIQ 2–8) et la FEVG médiane de 60 % (EIQ 56–63). Le |SLG| moyen était de 17,3 % ± 3,4 (IC 95 % 16,6–17,9). Le SLG était altéré chez 57 patients (54,3 % ; IC 95 % 44,8–63,5). Chez 30 patients tirés au sort, les coefficients de corrélation intraclasse intra- et inter-observateur étaient de 0,973 (IC 95 % 0,943–0,987) et 0,934 (IC 95 % 0,866–0,968), avec un accord sur le classement altéré/normal de κ = 0,93 et κ = 0,78 respectivement. La classification a identifié trois phénotypes ; le plus petit (n = 11) associait sexe féminin, mauvaise observance, hypertension sévère, hypertrophie ventriculaire gauche et dysfonction diastolique de grade II, avait la plus longue durée d’hypertension (médiane 7 ans contre 2 à 3 ans) et présentait le |SLG| moyen le plus bas (14,7 % contre 17,6 % et 17,5 % ; p = 0,029). La FEVG expliquait 17,8 % de la variance du |SLG| (β 0,27 par point de FEVG ; IC 95 % 0,16–0,38 ; p < 0,001). Après ajustement, l’âge (rapport de prévalence ajusté [RPa] 1,020 par année ; IC 95 % 1,004–1,036), l’indice de masse corporelle (RPa 1,040 par kg/m² ; IC 95 % 1,004–1,077), la pression artérielle systolique (RPa 1,12 par 10 mmHg ; IC 95 % 1,03–1,21) et l’hypertrophie ventriculaire gauche (RPa 1,45 ; IC 95 % 1,03–2,05) étaient associés à l’altération du SLG.
Conclusion : Plus de la moitié de ces patients hypertendus traités présentaient une altération de la fonction longitudinale malgré une FEVG préservée. L’âge, l’adiposité, la pression artérielle systolique et l’hypertrophie ventriculaire gauche étaient indépendamment associés à cette altération, avec des effets modestes qui doivent être considérés comme générateurs d’hypothèses.
MOTS CLÉS
Hypertension artérielle ; strain longitudinal global ; dysfonction ventriculaire gauche ; Cameroun.
SUMMARY
Background. Left ventricular ejection fraction (LVEF) remains within reference limits long after hypertensive myocardial injury has begun. Global longitudinal strain (GLS) detects that injury sooner, because subendocardial longitudinal fibres fail first. Strain data from sub-Saharan Africa remain limited, with recent reports from Cameroon and Togo, and none from the North of Cameroon. This study aimed to assess the prevalence, phenotypes and correlates of subclinical left ventricular dysfunction in a group of hypertensive outpatients in North Cameroon.
Methods. We conducted a cross-sectional analytical study at the cardiology department of the Garoua General Hospital between May and September 2026. Consecutive adults aged 21 years or over, treated for essential hypertension, with LVEF ≥ 50% and no symptoms or signs of heart failure, were enrolled. GLS was measured by two-dimensional speckle tracking across 17 segments and expressed as an absolute value; impairment was defined as |GLS| below 18%. Intra- and inter-observer reproducibility was assessed in 30 randomly selected patients with intraclass correlation coefficients, Bland–Altman analysis, coefficients of variation and Cohen’s kappa for the impaired/normal classification. Prevalence ratios were estimated by Firth-penalised Poisson regression with robust variance and small-sample correction, adjusting for covariates selected a priori on clinical grounds. Clinical phenotypes were derived by multiple correspondence analysis followed by Ward hierarchical clustering. The LVEF–GLS relation was modelled by linear regression with heteroskedasticity-consistent (HC3) standard errors.
Results. Of 119 patients assessed, 105 were analysed. Mean age was 56.4 years (SD 10.5) and 58 participants (55.2%) were women. Median duration of hypertension was 3 years (IQR 2–8) and median LVEF was 60% (IQR 56–63). Mean |GLS| was 17.3% (SD 3.4; 95% CI 16.6–17.9). GLS was impaired in 57 patients (54.3%, 95% CI 44.8–63.5). In 30 randomly selected patients, intra-observer and inter-observer intraclass correlation coefficients for GLS were 0.973 (95% CI 0.943–0.987) and 0.934 (95% CI 0.866–0.968), with agreement on the impaired/normal classification of κ = 0.93 and κ = 0.78 respectively. Clustering yielded three phenotypes; the smallest (n = 11) combined female sex, poor adherence, severe hypertension, left ventricular hypertrophy and grade II diastolic dysfunction, had the longest duration of hypertension (median 7 versus 2–3 years) and carried the lowest mean |GLS| (14.7% versus 17.6% and 17.5%; p = 0.029). LVEF accounted for 17.8% of the variance in |GLS| (β 0.27 per 1% LVEF, 95% CI 0.16–0.38; p < 0.001). After adjustment, age (adjusted prevalence ratio 1.020 per year, 95% CI 1.004–1.036), body mass index (1.040 per kg/m², 95% CI 1.004–1.077), systolic blood pressure (1.12 per 10 mmHg, 95% CI 1.03–1.21) and left ventricular hypertrophy (1.45, 95% CI 1.03–2.05) were associated with impaired GLS.
Conclusion. More than half of these treated hypertensive outpatients had impaired longitudinal function despite a preserved ejection fraction. Age, adiposity, systolic blood pressure and hypertrophy were independently associated with impairment, with modest effect sizes that should be regarded as hypothesis-generating.
KEY WORDS
Hypertension; global longitudinal strain; left ventricular dysfunction; Cameroon.
1. Cardiology Unit, Garoua General Hospital, Garoua, Cameroon
2. Faculty of Medicine and Biomedical Sciences, University of Yaoundé I, Yaoundé, Cameroon
3. Cardiology Unit, Yaoundé University Teaching Hospital, Yaoundé, Cameroon
Adresse pour correspondance
Prof. Hamadou Ba
Associate Professor at the Faculty of Medicine and Biomedical Sciences, University of Yaoundé I, Yaoundé, Cameroon.
Phone: (+237) 696 416 842
Email: This email address is being protected from spambots. You need JavaScript enabled to view it.
INTRODUCTION
Hypertension affects roughly 1.3 billion adults aged 30 to 79 years, and prevalence has doubled since 1990 [1]. The African region carries the heaviest share, and control rates there remain the lowest recorded [1,2]. In Cameroon, pooled estimates place prevalence at 32.1% for 2011–2018, up from 29.6% two decades earlier [3]. Hypertensive heart disease follows a recognised sequence: pressure overload, concentric remodelling, hypertrophy, impaired relaxation, and eventually overt heart failure. What that sequence conceals is the point at which contractile injury actually begins.
Left ventricular ejection fraction (LVEF) is a poor sentinel for that beginning. It reflects radial and circumferential shortening, which the mid-wall and epicardial fibres sustain, and it is preserved by compensatory thickening even when longitudinal function has already declined [4,5]. Longitudinal fibres run subendocardially. They are the first to suffer from raised wall stress, microvascular rarefaction and interstitial fibrosis [6]. Ejection fraction is also load-dependent and its reproducibility by two-dimensional Simpson biplane methods is modest [7].
Two-dimensional speckle tracking measures myocardial deformation directly. Global longitudinal strain (GLS) is more reproducible than ejection fraction, is less sensitive to loading conditions, and predicts outcome across ischaemic heart disease, valve disease, cardiomyopathy and cardiotoxicity [8–11]. In hypertension specifically, reduced GLS with preserved ejection fraction marks a stage of target-organ damage that conventional echocardiography misses [12,13]. A strain risk score has been shown to predict events in asymptomatic hypertensive heart disease [14].
Despite this, strain has not entered routine practice in most African echocardiography laboratories, and the published African literature, although growing, remains limited. Menanga and colleagues reported impaired GLS in 67% of 30 Cameroonian patients with heart failure and preserved ejection fraction [15], but those patients were already symptomatic. In Togo, Pessinaba and colleagues recently reported reduced GLS in 41.4% of 215 hypertensive patients with an ejection fraction above 50%, and identified duration of hypertension beyond ten years, dyslipidaemia, diabetes, obesity, concentric remodelling and concentric hypertrophy as associated factors [16]. That series is the closest published comparator to ours, but it used a threshold of 17% rather than 18%, expressed its associations as odds ratios for an outcome affecting more than 40% of patients, did not require the absence of symptoms or signs of heart failure, and recruited a younger population (mean age 46 years). What proportion of asymptomatic, treated hypertensive outpatients already carry subclinical contractile injury is therefore not known for this setting, nor is it known which clinical features identify them, nor how far ejection fraction can be relied upon to exclude such injury once it is within reference limits.
Garoua lies in the Sahelian north of Cameroon, a region with a distinct demographic and nutritional profile from the forested south where most national cardiovascular data originate. We set out to i) measure the prevalence of subclinical left ventricular systolic dysfunction by GLS in hypertensive outpatients with preserved ejection fraction, ii) describe the clinical phenotypes among which it clusters, iii) quantify the relation between ejection fraction and GLS, and iv) identify factors independently associated with impairment.
METHODS
Design and setting
This was a cross-sectional analytical study with prospective data collection, carried out in the cardiology outpatient unit of Garoua General Hospital, a tertiary referral centre and the highest level of the Cameroonian health system in the North Region. The study was designed in October 2025 and completed in September 2026. Recruitment and the reproducibility measurements were carried out between May and September 2026, after ethical approval had been granted on 13 May 2026.
Participants
We used exhaustive consecutive sampling: every eligible patient attending during the study period was invited to take part. Inclusion required an age of at least 21 years, representing the age of majority in Cameroon, a follow-up for essential hypertension of any duration, an LVEF of 50% or more on transthoracic echocardiography, and written informed consent.
Patients were not included if they had symptoms or signs of heart failure; an acute coronary syndrome within the preceding six months or documented coronary artery disease; moderate-to-severe valve disease; atrial fibrillation or atrial flutter; a permanent pacemaker or implantable cardioverter-defibrillator; secondary hypertension; or ongoing pregnancy. Patients with poor echogenicity not allowing an accurate measurement of GLS were excluded.
Clinical assessment
A structured interview recorded age, sex, occupation, educational level, marital status and place of residence. Comorbidities were documented, including diabetes, smoking with pack-year calculation, alcohol use, dyslipidaemia and hyperuricaemia. For hypertension we recorded date of diagnosis, current antihypertensive regimen and adherence.
Blood pressure was measured using an M7 intelli IT Afib (HEM-7380T1-EBK) automated blood pressure machine from OMRON Healthcare (Kyoto 617-0002, Japan), with an appropriately sized cuff, in a seated position, and after at least five minutes of rest. Two BP measurements were performed and the mean value was calculated and recorded. Heart rate and oxygen saturation were recorded. Weight and height were measured, and body mass index and body surface area calculated. Cardiovascular examination sought signs of right and left ventricular failure, and any patient with such signs was not included.
Blood pressure control was graded from office readings: controlled (< 140/90 mmHg), uncontrolled grade 1 (140–159 and/or 90–99 mmHg), uncontrolled grade 2 (160–179 and/or 100–109 mmHg), and severe hypertension (≥ 180 and/or ≥ 110 mmHg) [17]. Adherence was assessed with the six-item Girerd questionnaire and classified as good (score 0), minor problem (score 1–2) or poor adherence (score ≥ 3) [18].
Diabetes was defined as a documented diagnosis under follow-up. Dyslipidaemia was defined as total cholesterol above 2 g/L (5.17 mmol/L), LDL cholesterol above the threshold set by the patient’s global cardiovascular risk, or current lipid-lowering treatment [19]. Obesity was defined as body mass index 30 kg/m² or more.
Echocardiography
We used a Vivid™ T9 from GE Healthcare (Chicago, IL 60661, USA) ultrasound machine with a 3Sc-RS transducer. Studies were performed with the patient in the left lateral decubitus position. From the parasternal long-axis view we measured interventricular septal thickness, left ventricular internal diameter and posterior wall thickness at end-diastole, from which left ventricular mass was derived and indexed to body surface area. Relative wall thickness was calculated. Geometry was classified into four categories following the American Society of Echocardiography and European Association of Cardiovascular Imaging recommendations: normal geometry, concentric remodelling, concentric hypertrophy and eccentric hypertrophy, using indexed mass thresholds of 95 g/m² in women and 115 g/m² in men and a relative wall thickness threshold of 0.42 [20]. Left ventricular hypertrophy was defined as indexed mass above those thresholds, whatever the relative wall thickness.
Ejection fraction was obtained by the biplane method of discs from apical four- and two-chamber views, with manual endocardial tracing from the septal to the lateral annulus.
Diastolic function was assessed from transmitral pulsed Doppler (E and A velocities, E/A ratio, E-wave deceleration time), tissue Doppler at the septal and lateral mitral annulus (e′ velocities, mean E/e′), left atrial volume indexed by the biplane method, and peak tricuspid regurgitant velocity. Grading followed the 2016 joint recommendations [21]. Elevated filling pressures were defined by mean E/e′ above 14, indexed left atrial volume above 34 mL/m², septal e′ below 7 cm/s and tricuspid regurgitant velocity above 2.8 m/s, with two or more positive criteria taken as elevated.
Right ventricular systolic function was assessed by tricuspid annular plane systolic excursion (abnormal below 17 mm) and by peak systolic tissue Doppler velocity at the lateral tricuspid annulus (abnormal below 9.5 cm/s).
Strain analysis
Images for speckle tracking were acquired at 70 to 100 frames per second at end-expiration from apical long-axis, two-, three- and four-chamber views. Analysis used automated function imaging. Three anchor points were placed manually (two basal and one apical) after which the software tracked the endocardial border through the cardiac cycle. Tracking quality was inspected and the region of interest adjusted where necessary before the operator validated the analysis. The left ventricle was divided into 17 segments and segmental longitudinal strain values generated for each apical view.
GLS is reported throughout as an absolute value, to avoid the sign confusion that attends negative strain [20]. Impairment was graded as normal (|GLS| ≥ 18%), mildly reduced (16–17.9%), moderately reduced (14–15.9%) and severely reduced (< 14%). For all binary analyses, subclinical systolic dysfunction was defined as |GLS| below 18%, a value close to the lower limit of the reference range reported in meta-analyses of healthy subjects [22,23].
Strain analysis was performed offline with EchoPAC software (GE Vingmed Ultrasound, Horten, Norway) version 204.22.1. Because GLS values are not fully interchangeable across vendors and software versions, the same ultrasound system and software were used for all patients [24]. Intra- and inter-observer reproducibility of GLS was assessed in 30 patients drawn at random from the 105 analysed, using a computer-generated random sequence. For each of them three measurements of GLS were available: the initial measurement used in the main analysis, a second measurement obtained by the same observer at least two weeks later, blinded to the first result and to clinical data, and a third obtained independently by a second experienced observer blinded to both previous measurements and to clinical data. Intra-observer agreement was assessed by comparing the first two measurements and inter-observer agreement by comparing the first and third.
Statistical analysis
Analyses were performed in SPSS version 27 and verified independently in Python 3.13 (statsmodels, prince, scipy); Firth-penalised Poisson regression with Morel–Bokossa–Neerchal corrected variance was implemented in Python. Continuous variables are summarised as mean with standard deviation when their distribution was normal, and as median with interquartile range otherwise; normality was assessed with the Shapiro–Wilk test complemented by visual inspection of histograms and Q–Q plots. Categorical variables are presented as counts and percentages. Between-group comparisons of continuous variables used one-way ANOVA for normally distributed variables and the Kruskal–Wallis test otherwise; categorical variables were compared with the chi-square test, or Fisher's exact test when expected cell counts were below 5. The 95% confidence interval for the prevalence of impaired GLS was calculated by the Wilson score method.
Phenotyping
Multiple correspondence analysis (MCA) was performed on eight active categorical variables: sex, diabetes, dyslipidaemia, physical inactivity, blood pressure control, adherence, diastolic function grade and left ventricular hypertrophy. GLS status, body mass index category, duration of hypertension and age category were projected as supplementary variables, so that they informed interpretation without shaping the factorial axes. This separation matters: had GLS helped build the axes, any subsequent association between cluster and GLS would have been circular. Ward hierarchical clustering [25] was then applied to the factorial coordinates of the first two dimensions, weighted by the square root of the corresponding eigenvalues. The number of classes was chosen by inspection of the dendrogram and of the gains in inertia, together with the clinical interpretability of the resulting classes. The robustness of that choice was checked afterwards with internal validity indices (average silhouette width and the ratio of between-class to total inertia) and with a bootstrap stability analysis; these did not guide the choice and are reported in Supplementary Material S1. We used the ANOVA and Kruskal–Wallis tests to compare the distribution across the clusters.
LVEF–GLS relation
Normality of LVEF and |GLS| was checked by Kolmogorov–Smirnov and Shapiro–Wilk tests [26]. Linearity was tested by the Box–Tidwell procedure [27]. Because homoskedasticity could not be assumed and influential observations were present, the model was fitted with heteroskedasticity-consistent HC3 standard errors.
Associated factors
Because impaired GLS was common, odds ratios from logistic regression would have overstated the prevalence ratio. We therefore used Poisson regression with robust (sandwich) variance estimation, which returns prevalence ratios directly [28,29]. The multivariable model was restricted to six covariates selected a priori on clinical grounds as established determinants of longitudinal strain in hypertension, without reference to the univariable results: age, sex, body mass index, systolic blood pressure, diabetes and left ventricular hypertrophy [30]. One indicator was retained per domain, so that redundant structural and haemodynamic markers (relative wall thickness, left atrial dimensions, diastolic function grade and filling pressures) were not entered together; with 57 events, this corresponds to 9.5 events per parameter. Collinearity was assessed with variance inflation factors. All crude and adjusted models were fitted with Firth-type penalised modified Poisson regression, which reduces small-sample bias and yields finite estimates under complete separation [31] and Morel–Bokossa–Neerchal small-sample correction with robust variances were computed, which keeps the variance positive definite when an exposure category is perfectly predicted [32]; this was the case for elevated filling pressures, since all affected patients had impaired GLS. In sensitivity analyses, poor adherence or duration of hypertension was added to the model, and the main model was refitted by conventional unpenalised robust Poisson regression. Adjusted prevalence ratios are reported with 95% confidence intervals. All tests were two-sided and the significance threshold was 5%.
Reproducibility
Intra- and inter-observer agreement for GLS was quantified with intraclass correlation coefficients for absolute agreement, single measures (two-way mixed-effects model for intra-observer and two-way random-effects model for inter-observer agreement), with 95% confidence intervals derived from the F distribution; values above 0.90 were taken to indicate excellent reliability and values from 0.75 to 0.90 good reliability [33]. Systematic error was assessed by Bland–Altman analysis, the mean of the paired differences giving the bias, tested against zero by a paired t test, and the 95% limits of agreement being the bias plus or minus 1.96 times the standard deviation of the differences [34]. Random error was expressed as the within-subject coefficient of variation, computed as the standard deviation of the differences divided by the square root of two and by the mean of the paired measurements, and as the repeatability coefficient (1.96 times the standard deviation of the differences), which gives the difference expected between two measurements of the same patient in 95% of cases. Because the clinical inference rests on a threshold, agreement on the binary classification of GLS as impaired (|GLS| < 18%) was quantified separately with Cohen’s kappa, for which 95% confidence intervals were obtained by percentile bootstrap with 20,000 resamples, and with the observed proportion of concordant classifications; kappa values above 0.80 were taken to indicate almost perfect agreement and values from 0.61 to 0.80 substantial agreement.
Ethics
The study followed the Declaration of Helsinki and the 2016 CIOMS international ethical guidelines. Ethical approval was granted by the Regional Ethics Committee for Human Health Research of the North Region (N°208/CERSH/NO/2026) (13 May 2026). Institutional approval was also obtained from the Faculty of Medicine and Biomedical Sciences, University of Yaoundé I (N°0495/UY1/FMSB/VDRC/DAARS/CSD/CSDA) (2 September 2026). Administrative authorisation was obtained from the management of Garoua General Hospital. All participants gave written informed consent. Study investigations were provided free of charge.
RESULTS
Participants and Characteristics
Of 119 patients assessed, seven were not included: four with atrial fibrillation, one with a recent acute coronary syndrome and two with an ejection fraction below 50%. Seven others were subsequently excluded because of poor echogenicity. The analysis therefore rests on 105 patients, with complete data for all modelled variables except duration of hypertension, which was missing for one patient (Figure 1).
Mean age was 56.4 years (SD 10.5, range 33–84) and 58 participants (55.2%) were women (Table 1). Secondary education was the commonest attainment (43 patients, 41.0%), while 11 (10.5%) had never attended school. Most participants lived with a partner (82, 78.1%).
Median duration of hypertension was 3 years (IQR 2–8). Mean systolic pressure at the study visit was 144.9 mmHg (SD 24.1) and mean diastolic pressure 90.3 mmHg (SD 13.8). Blood pressure was controlled in 39 patients (37.1%); 45 (42.9%) had uncontrolled grade 1 hypertension, 15 (14.3%) grade 2, and 6 (5.7%) severe hypertension.
Dual therapy was the commonest regimen (66 patients, 62.9%), followed by monotherapy (29, 27.6%), triple therapy (9, 8.6%) and four or more agents (1, 1.0%). Among the 76 patients on combination therapy, 59 (77.6%) received a fixed-dose combination. Calcium channel blockers were the most prescribed class (79, 75.2%), then angiotensin-converting enzyme inhibitors (47, 44.8%), diuretics (32, 30.5%), beta-blockers (18, 17.1%) and angiotensin receptor blockers (14, 13.3%). Twenty patients (19.0%) were on a statin and 15 (14.3%) on aspirin.
Adherence was good in 24 patients (22.9%), showed a minor problem in 48 (45.7%) and was poor in 33 (31.4%). Physical inactivity was the commonest comorbidity (67, 63.8%), ahead of diabetes (34, 32.4%) and dyslipidaemia (32, 30.5%). Mean body mass index was 28.3 kg/m² (SD 4.8); 45 patients (42.9%) were overweight and 33 (31.4%) obese. Only one participant reported current or past smoking.
Echocardiographic findings
Median LVEF was 60% (IQR 56–63) and median indexed left ventricular mass 81 g/m² (IQR 69–102), with a median relative wall thickness of 0.43 (IQR 0.39–0.47) (Table 2). Abnormal geometry was present in 67 patients (63.8%), dominated by concentric remodelling (46, 43.8%); concentric hypertrophy affected 15 (14.3%) and eccentric hypertrophy 6 (5.7%). Left ventricular hypertrophy, on indexed mass criteria, was present in 21 patients (20.0%).
Median indexed left atrial volume was 24.6 mL/m² (IQR 20.0–28.9) and median mean E/e′ was 7.6 (IQR 6.2–10.1). Diastolic function was normal in 46 patients (43.8%); 52 (49.5%) had grade I dysfunction, 6 (5.7%) grade II and 1 (1.0%) grade III. Filling pressures were elevated in 6 patients (5.7%). Right ventricular function was preserved in all participants, with a median tricuspid annular plane systolic excursion of 24 mm (IQR 22–27).
Mean |GLS| was 17.3% (SD 3.4; 95% CI 16.6–17.9), with values spanning 8.4% to 25.9%. GLS was impaired in 57 patients, a prevalence of 54.3% (95% CI 44.8–63.5). Impairment was mild in 18 (17.1%), moderate in 18 (17.1%) and severe in 21 (20.0%).
Reproducibility was assessed in 30 randomly selected patients, whose GLS spanned the range of the cohort (8.4% to 23.5%; mean 16.3%). Intra-observer agreement was excellent: the intraclass correlation coefficient was 0.973 (95% CI 0.943–0.987), the bias was −0.08% (95% CI −0.42 to +0.25; p = 0.615) with limits of agreement of −1.84% to +1.68%, the within-subject coefficient of variation was 3.9% and the repeatability coefficient 1.8%. Inter-observer agreement was also excellent but less tight: the intraclass correlation coefficient was 0.934 (95% CI 0.866–0.968), the bias was +0.09% (95% CI −0.42 to +0.59; p = 0.727) with limits of agreement of −2.55% to +2.72%, the within-subject coefficient of variation was 5.9% and the repeatability coefficient 2.6%. Neither bias differed from zero, so no systematic drift was detected between readings or between observers; the wider limits of agreement between observers indicate greater random error. Classification of GLS as impaired was reproduced in 29 of 30 patients within the observer (κ = 0.93, 95% CI 0.76–1.00) and in 27 of 30 patients between observers (κ = 0.78, 95% CI 0.53–1.00). All four discordant classifications concerned patients whose GLS lay within 2.5 percentage points of the 18% threshold (18.4% to 20.5% at the first reading); no patient was reclassified when GLS was far from the threshold.
Clinical phenotypes
The first two MCA dimensions accounted for 14.8% and 12.5% of raw inertia respectively (27.3% cumulative), which is typical for multiple correspondence analysis of binary and short ordinal variables, where raw inertia is structurally deflated. Applying the Benzécri correction, the same two dimensions accounted for 84.5% of adjusted inertia, supporting their retention.
Dimension 1 separated a favourable profile — male sex, physical activity, good adherence, controlled or grade 1 hypertension, absent hypertrophy — from a profile of advanced hypertensive heart disease marked by hypertrophy, poor adherence and higher-grade diastolic dysfunction. Among the supplementary variables, all of which lay close to the origin, normal GLS, overweight, obesity, age over 55 years and hypertension of ten years or less fell on the favourable side of dimension 1, whereas impaired GLS, normal body mass index, age 55 years or under and hypertension of more than ten years fell on the unfavourable side. Dimension 2 discriminated chiefly among the uncontrolled categories. The category map is shown in Figure 2.
Ward clustering on the weighted factorial coordinates was cut at three classes (Figure 3). The largest gain in inertia occurred at the final fusion (linkage distance 2.84, against 2.06 for the preceding one), which would have separated phenotype 3 from all other patients; the three-class partition was retained because the preceding fusion divided the remaining 94 patients into two groups with clinically distinct profiles, so the choice rests on interpretability as well as on the dendrogram. Internal validity indices confirmed that two classes gave the strongest separation (average silhouette width 0.587 against 0.431 for three classes), while three classes accounted for 59.3% of total inertia against 38.9%. On bootstrap resampling, phenotypes 1 and 3 were reproduced with mean Jaccard coefficients of 0.78 and 0.73, whereas phenotype 2 reached 0.56 (Supplementary Material S1). Class profiles are given in Table 3 and the projection of individuals in Figure 4.
Phenotype 1 (n = 63) was the largest and most homogeneous. Women predominated (57%), physical inactivity was near-universal (75%), grade 1 uncontrolled hypertension was the rule (63%) and adherence showed a minor problem in 56%. Grade I diastolic dysfunction was present in 57%, but hypertrophy was almost absent (97% without). These patients were older (52% over 55 years) and mostly overweight. GLS was impaired in 33 of 63 (52.4%).
Phenotype 2 (n = 31) was male-predominant (58%), younger (58% aged 55 years or under), physically active (61%) and the best adherent (45% good adherence). Hypertension was less well controlled — grade 2 in 48% — yet structural involvement was limited, with 61% free of hypertrophy. GLS was impaired in 15 of 31 (48.4%). This group therefore combined worse office blood pressure with less myocardial consequence. Its median duration of hypertension was the shortest (2 years, IQR 1–7), but the proportion with more than ten years of hypertension was the same as in phenotype 1 (13%), so duration alone does not explain this dissociation.
Phenotype 3 (n = 11) was the smallest and the most distinct, fusing late with the rest of the partition. Women made up 82%, adherence was poor in 82%, hypertension was severe in 55%, hypertrophy was present in 64% and diastolic dysfunction reached grade II in 55%. GLS was impaired in 9 of 11 (81.8%). This group also had the longest exposure: median duration of hypertension was 7 years (IQR 4–10), against 3 years in phenotype 1 and 2 years in phenotype 2, and 27% had had hypertension for more than ten years, against 13% in each of the other phenotypes (Kruskal–Wallis p = 0.093). Body mass index was mostly normal (55%), so its severity is not attributable to adiposity.
Mean |GLS| differed across the three phenotypes: 17.6% (SD 2.9), 17.5% (SD 3.8) and 14.7% (SD 4.0) respectively (one-way ANOVA p = 0.029; Kruskal–Wallis p = 0.031) (Figure 5). The corresponding comparison using the dichotomised variable did not reach significance (chi-square p = 0.14), an expected consequence of the information lost on dichotomisation.
Relation between ejection fraction and strain
Neither LVEF nor |GLS| departed from normality (Shapiro–Wilk p = 0.076 and p = 0.50; Kolmogorov–Smirnov p = 0.118 and p = 0.200). The Box–Tidwell interaction term was not significant (p = 0.231), so the relation was treated as linear.
LVEF was positively associated with |GLS|: β = 0.268 per 1% of LVEF (HC3 standard error 0.054; 95% CI 0.161–0.376; p < 0.001), giving the fitted line |GLS| = 1.30 + 0.268 × LVEF. The Pearson correlation was 0.42 (p < 0.001) and the coefficient of determination 0.178. In other words, ejection fraction accounted for 17.8% of the variance in longitudinal strain, and scatter about the regression line was wide. Median LVEF declined across the first three strain categories — 61%, 59% and 56% — and did not decline further in the severely reduced category (57%); the ranges overlapped throughout (Figure 6).
Factors associated with impaired strain
In univariable analysis (Table 4), age was associated with impaired GLS (crude prevalence ratio 1.016 per year, 95% CI 1.001–1.032; p = 0.040). Systolic blood pressure, analysed as a continuous variable, was also associated with impairment (1.10 per 10 mmHg, 95% CI 1.03–1.18; p = 0.004). Among blood pressure categories, only severe hypertension reached significance (1.74, 95% CI 1.07–2.84; p = 0.025); grades 1 and 2 did not. Poor adherence showed only a trend towards impairment (1.72, 95% CI 0.98–3.03; p = 0.060). Sex, duration of hypertension, diabetes, dyslipidaemia, physical inactivity and body mass index showed no significant crude association, though body mass index and dyslipidaemia trended.
Among echocardiographic variables, left ventricular hypertrophy was associated with impaired GLS (1.59, 95% CI 1.14–2.21; p = 0.006), as were left atrial area (1.04 per cm², p = 0.008) and indexed left atrial volume (1.02 per mL/m², p = 0.009). All six patients with elevated filling pressures had impaired GLS. With conventional robust variance, this complete separation produces an artificially narrow interval (1.94, 95% CI 1.60–2.35); the penalised estimate with small-sample corrected variance was 2.08 (95% CI 1.67–2.60; p < 0.001), but it rests on six patients and is not interpreted further. Relative wall thickness was also associated with impairment. Because it is expressed on a 0–1 scale, it was reparameterised per 0.10 unit, which gives a prevalence ratio of 1.35 (95% CI 1.02–1.80; p = 0.038) rather than the numerically unstable 20.7 (95% CI 1.19–360.4) obtained per whole unit. Systolic pulmonary artery pressure was not associated (p = 0.658). Diastolic dysfunction trended without reaching threshold (1.43, 95% CI 0.97–2.10; p = 0.072).
In the multivariable model with the six covariates selected a priori (n = 105; 57 events), variance inflation factors ranged from 1.02 to 1.08. Four variables were independently associated with impaired GLS (Table 4): age (adjusted prevalence ratio 1.020 per year, 95% CI 1.004–1.036; p = 0.015), body mass index (1.040 per kg/m², 95% CI 1.004–1.077; p = 0.029), systolic blood pressure (1.12 per 10 mmHg, 95% CI 1.03–1.21; p = 0.006) and left ventricular hypertrophy (1.45, 95% CI 1.03–2.05; p = 0.033). Sex (0.82, 95% CI 0.58–1.16) and diabetes (1.08, 95% CI 0.75–1.55) were not. Each additional year of age was associated with a 2% higher prevalence of impaired GLS, each additional kg/m² with a 4% higher prevalence and each 10 mmHg of systolic pressure with a 12% higher prevalence. The associations with age, body mass index and systolic pressure persisted when poor adherence or duration of hypertension was added and with conventional unpenalised robust Poisson regression; that with hypertrophy was less stable (p = 0.054 after adding duration of hypertension, n = 104).
DISCUSSION
Among 105 treated hypertensive outpatients whose ejection fraction was normal and who had no symptoms of heart failure, 54.3% already had impaired longitudinal strain, and one in five had severe impairment. Ejection fraction explained under a fifth of the variance in strain. After adjustment for covariates selected a priori, age, body mass index, systolic blood pressure and hypertrophy were independently associated with impaired strain. Three clinical phenotypes emerged, of which the smallest (women with poor adherence, severe hypertension, hypertrophy and the longest duration of hypertension) carried by far the heaviest strain burden.
Prevalence
Our figure of 54.3% sits within the range reported for hypertensive cohorts with preserved ejection fraction. Taleb Bendiab and colleagues found GLS reduction in 45.5% of Algerian hypertensive patients with normal ejection fraction, with marked reduction in 11% [13,35]. Ferruzzi and colleagues reported that impaired GLS in hypertensive patients was associated with mild cognitive impairment, one expression of the target-organ involvement that accompanies subclinical dysfunction [36]. Menanga and colleagues reported 67% in Cameroonian patients, but their cohort had established heart failure with preserved ejection fraction, so a higher figure is expected [15]; that our asymptomatic patients fall below theirs, and above the Algerian outpatient figure, is internally coherent. The closest comparison is the Togolese series of Pessinaba and colleagues, who studied 215 hypertensive patients with an ejection fraction above 50% and found reduced GLS in 41.4% of them, using a threshold of 17% [16]. Applying that same threshold to our cohort gives 41.0% (95% CI 32.0–50.5), and their mean GLS, 17.4% (SD 4.0), is close to ours, 17.3% (SD 3.4). Two West African series recruited independently, in different countries and with different age structures, therefore agree closely once the threshold is harmonised. This convergence argues that our estimate is not an artefact of local recruitment, and it shows how much of the apparent variation between published prevalences reflects the choice of cut-off rather than a difference in myocardial disease.
Three features of our cohort help explain a prevalence at the upper end of the outpatient range. Blood pressure was uncontrolled in 62.9%. Adherence was good in fewer than a quarter. Adiposity was near-universal, with 74.3% overweight or obese. Uncontrolled blood pressure has been linked to GLS decline in hypertensive patients [13], hypertension more generally to reduced longitudinal strain [37], and obesity to adverse changes in cardiac morphology and ventricular function [38]; poor adherence has not been studied directly in this respect, but plausibly acts through a greater cumulative pressure load.
The comparison across studies is nonetheless imperfect, because thresholds differ. We used |GLS| below 18%, consistent with the lower limit of normal in the chamber quantification update [20]; others have used 16% or 17%. Vendor differences in tracking algorithms add further variation, which the standardisation initiative sought to narrow but has not abolished [24]. Prevalence estimates in this field should be read as vendor- and threshold-conditional.
What ejection fraction does not tell us
The regression of |GLS| on LVEF is, to our mind, the most clinically consequential result. The association was positive and statistically strong (p < 0.001), yet the coefficient of determination was 0.178. Four-fifths of the variance in longitudinal strain lay outside anything ejection fraction captured.
Benyounes and colleagues reported r = −0.53 across an unrestricted range of ejection fraction in routine practice [39]. Our weaker correlation (r = 0.42 on absolute values) is what restriction of range predicts: by design, every patient here had an ejection fraction of at least 50%, which truncates the very variability that drives the correlation upward in unselected series. The implication is not that the two indices disagree, but that within the preserved range (precisely the range in which early detection matters) ejection fraction carries little information about longitudinal mechanics. Median ejection fraction did decline across our strain categories, from 61% to 57%, but the distributions overlapped so heavily that no ejection fraction cut-off could have separated them.
This is a practical argument for acquiring strain rather than inferring it. A patient with an ejection fraction of 60% may have a GLS of 21% or of 12% as shown in figure 6; nothing in the Simpson biplane measurement distinguishes them.
Phenotypes
The three-class solution describes a severity gradient rather than three unrelated syndromes, and its interest lies chiefly in phenotype 3. Those eleven patients had severe hypertension, poor adherence, hypertrophy and grade II diastolic dysfunction, with impaired strain in 82% and a mean |GLS| of 14.7%. They also had the longest recorded exposure, with a median duration of hypertension of 7 years and 27% beyond ten years, whereas 55% had a normal body mass index. Cumulative exposure is therefore a plausible contributor to their severity; adiposity is not.
Two readings, not mutually exclusive, are available. The first is haemodynamic: a longer duration combined with poor adherence in 82% implies a greater cumulative pressure load than in the other phenotypes, since intermittently treated hypertension adds to the burden that recorded duration already reflects. The difference in duration did not reach statistical significance (p = 0.093), however, and eleven patients cannot separate these two components. The second is that some patients carry a myocardial susceptibility that is not captured by the variables we measured. Female predominance in this class (82%), against 55% in the cohort, is consistent with reports of greater concentric remodelling and steeper strain decline in hypertensive women, though our numbers are far too small to make that claim. Eleven patients cannot support a phenotype in the strong sense. We report the class because its separation on dimension 1 was clean and its fusion with the rest of the partition late, but it needs replication.
Phenotype 2 is instructive in the opposite direction. These patients were younger, more active, better adherent and had worse office blood pressure (grade 2 in 48%) yet the least myocardial involvement, with normal GLS in 52%. Blood pressure at a point in time and myocardial consequence are not the same quantity, and the dissociation between them is exactly what strain imaging is for.
We would emphasise that GLS was held out of the axis construction. Any association between class and strain therefore emerged from the clinical variables alone, which is what makes the gradient in mean |GLS| across classes (p = 0.029) worth reporting. That the same comparison lost significance on the dichotomised variable (p = 0.14) is a reminder of what dichotomisation costs.
Age, adiposity, blood pressure and hypertrophy
Four variables remained associated with impaired strain after adjustment, all with modest effect sizes: each year of age carried a 2% higher prevalence of impaired strain, each kg/m² of body mass index a 4% higher prevalence, each 10 mmHg of systolic pressure a 12% higher prevalence and hypertrophy a 45% higher prevalence. The lower confidence bounds for age, body mass index and hypertrophy lay close to unity, and the association with hypertrophy depended on model specification, so these estimates should be read as hypothesis-generating.
The direction of these findings is nonetheless expected. Longitudinal function declines with age in healthy populations as subendocardial fibres are lost and interstitial collagen accumulates [40]. Excess adiposity raises afterload and promotes myocardial inflammation and fibrosis, and obesity in early adulthood and middle age has been associated with impaired longitudinal strain in a population cohort [41]. Reduced GLS is a consistent finding in hypertension [35] and is more marked when left ventricular geometry is abnormal [37]; uncontrolled blood pressure and hypertrophy were likewise independently associated with GLS decline in Algerian hypertensive patients [13]. In Togo, obesity and concentric hypertrophy were likewise associated with reduced GLS [16], which points in the same direction in a West African population. Two of the Togolese associations were not reproduced here: duration of hypertension beyond ten years and dyslipidaemia, reported with odds ratios of 14.3 and 12.2 in Lomé, gave prevalence ratios of 1.18 (95% CI 0.75–1.84) and 1.34 (95% CI 0.95–1.90) in our data. Part of that gap is arithmetical, since odds ratios overstate prevalence ratios when the outcome affects more than 40% of patients, and part is structural, since only 14% of our patients had been hypertensive for more than ten years, which leaves little contrast on that variable.
Hypertrophy retained an association after adjustment, yet it affected only 21 patients whereas 57 had impaired strain, which is consistent with the premise that longitudinal impairment can precede structural change. Left atrial dimensions, relative wall thickness and diastolic function were not entered in the model, to avoid redundancy with hypertrophy, so their independent contribution was not assessed. Sex and diabetes were not associated with impaired strain; for diabetes in particular, this should not be read as evidence of no effect, since the estimate was imprecise (95% CI 0.75–1.55). Elevated filling pressures could not be assessed meaningfully: all six affected patients had impaired strain, and even the penalised estimate rests on too few patients to be interpreted.
Clinical implications
If more than half of asymptomatic, treated hypertensive outpatients have impaired longitudinal function, and if ejection fraction cannot identify them, the case for adding strain to the routine hypertensive echocardiogram needs to be considered on clinical grounds. The obstacle in this setting is not conceptual but material: strain requires adequate acoustic windows, a frame rate of 70 to 100 per second, vendor software and an operator trained to inspect tracking rather than accept it. Where those are available, the incremental time is a few minutes.
The immediate use of an abnormal GLS in a hypertensive patient is to intensify what is already indicated: blood pressure control, adherence support and weight management. Whether treatment guided by strain improves outcome in hypertension has not been tested, and we make no such claim.
Strengths and limitations
The main strengths are the consecutive enrolment of all eligible patients over five months, complete data for every modelled variable except one missing duration of hypertension, the use of Poisson regression with robust variance rather than logistic regression – which would have inflated the effect estimates given a 54% prevalence – and the decision to hold GLS out of the correspondence analysis, which protects the phenotype–strain association from circularity.
The limitations are substantial and should temper interpretation.
The study was conducted at a single tertiary centre and sampled patients attending a cardiology clinic. Those patients are not representative of hypertensive adults in the North Region, and prevalence here should not be read as a population estimate. Referral to a cardiology service is itself a marker of complexity.
The design is cross-sectional. Nothing here establishes that age or adiposity cause strain impairment, and no prognostic inference is possible without follow-up. Furthermore, the study design did not permit us to distinguish participants with pre-existing alteration of GLS from those who developed it under treatment.
Reproducibility was assessed on a random subsample of 30 patients rather than on the whole cohort, and although agreement was excellent, the inter-observer limits of agreement of about ±2.6% are of the same order as the distance separating many patients from the 18% threshold: all four discordant classifications occurred in that zone, so prevalence estimates based on a single cut-off carry a measurement-related uncertainty that a single figure cannot convey. All measurements were also obtained with a single vendor’s system and software, and GLS values obtained with other vendors or software versions may differ, which further limits the transposition of our threshold-based prevalence estimate [24].
The multivariable model was limited to six covariates to match the number of events, at the cost of leaving out other plausible determinants such as adherence and duration of hypertension, which were examined only in sensitivity analyses. Although the covariates were selected on clinical grounds rather than from the data, the set was not registered before data collection, and a confirmatory study should pre-register it. Table 4 also reports eighteen crude comparisons with no adjustment for multiplicity, so crude p values close to 0.05 should be read with caution; the conclusions rest on the adjusted estimates of a single pre-specified model.
Phenotype 3 rests on eleven patients, and class membership at that size is sensitive to small changes in clustering parameters, so the profile should be treated as descriptive. Phenotype 2 was the least well separated and the least stable of the three classes on bootstrap resampling (Supplementary Material S1), so the boundary between phenotypes 1 and 2 should also be regarded as provisional.
Coronary disease was excluded on clinical grounds and history alone, without stress testing or angiography. Silent ischaemia in a cohort with 32% diabetes could contribute to regional strain reduction. Natriuretic peptides and troponin were not measured, so we cannot say how strain impairment relates to circulating markers of wall stress and myocyte injury in this population. Neither ambulatory nor home blood pressure monitoring was performed, so white-coat and masked hypertension cannot be separated from sustained hypertension, and the control categories rest on office readings at a single visit.
Finally, six patients had elevated filling pressures and one had grade III diastolic dysfunction. Categories of that size cannot support stable estimates, as the complete separation for elevated filling pressures and the very wide interval obtained for relative wall thickness per whole unit (1.19 to 360.4) illustrate.
CONCLUSION
Among treated hypertensive outpatients in northern Cameroon with a preserved ejection fraction and no heart failure symptoms, 54.3% had impaired global longitudinal strain and 20.0% had severe impairment. Ejection fraction explained 17.8% of the variance in strain, which is too little for it to serve as a surrogate within the preserved range. Age, body mass index, systolic blood pressure and hypertrophy were independently associated with impairment, with modest effect sizes. A minority phenotype (women with poor adherence, severe hypertension, hypertrophy, grade II diastolic dysfunction and the longest duration of hypertension) carried a disproportionate share of the strain burden, and merits study in a larger sample.
These findings support adding strain analysis to the hypertensive echocardiogram where the technology allows, and they identify blood pressure control, adherence support and weight management as the modifiable targets most plausibly linked to longitudinal function in this population. Prospective follow-up is needed to establish whether subclinical impairment detected this way predicts progression to overt heart failure, and whether treatment guided by it alters that course.
|
|
|
REFERENCES |
|
1. Zhou B, Carrillo-Larco RM, Danaei G, Riley LM, Paciorek CJ, Stevens GA, et al. Worldwide trends in hypertension prevalence and progress in treatment and control from 1990 to 2019: a pooled analysis of 1201 population-representative studies with 104 million participants. Lancet. 2021 ;398(10304):957–80. doi:10.1016/S0140-6736(21)01330-1 2. Mills KT, Stefanescu A, He J. The global epidemiology of hypertension. Nat Rev Nephrol. 2020;16(4):223–37. 3. Kuate Defo B, Mbanya JC, Kingue S, Tardif JC, Choukem SP, Perreault S, et al. Blood pressure and burden of hypertension in Cameroon, a microcosm of Africa: a systematic review and meta-analysis of population-based studies. J Hypertens. 2019 4. Bogaert J, Rademakers FE. Regional nonuniformity of normal adult human left ventricle. Am J Physiol Heart Circ Physiol. 2001;280(2):H610–20. 5. Braunwald E. Heart Failure. JACC Heart Fail. 2013;1(1):1–20. 6. Geyer H, Caracciolo G, Abe H, Wilansky S, Carerj S, Gentile F, et al. Assessment of Myocardial Mechanics Using Speckle Tracking Echocardiography: Fundamentals and Clinical Applications. J Am Soc Echocardiogr. 2010;23(4):351–69. 7. Thavendiranathan P, Grant AD, Negishi T, Plana JC, Popović ZB, Marwick TH. Reproducibility of Echocardiographic Techniques for Sequential Assessment of Left Ventricular Ejection Fraction and Volumes. J Am CollCardiol. 2013;61(1):77–84. 8. Antoni ML, Mollema SA, Delgado V, Atary JZ, Borleffs CJW, Boersma E, et al. Prognostic importance of strain and strain rate after acute myocardial infarction. Eur Heart J. 2010;31(13):1640–7. 9. Liu H, Pozios I, Haileselassie B, Nowbar A, Sorensen LL, Phillip S, et al. Role of Global Longitudinal Strain in Predicting Outcomes in Hypertrophic Cardiomyopathy. Am J Cardiol. 2017;120(4):670–5. 10. Magne J, Cosyns B, Popescu BA, Carstensen HG, Dahl J, Desai MY, et al. Distribution and Prognostic Significance of Left Ventricular Global Longitudinal Strain in Asymptomatic Significant Aortic Stenosis. JACC Cardiovasc Imaging. 2019;12(1):84–92. 11. Plana JC, Galderisi M, Barac A, Ewer MS, Ky B, Scherrer-Crosbie M, et al. Expert Consensus for Multimodality Imaging Evaluation of Adult Patients during and after Cancer Therapy: A Report from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. J Am Soc Echocardiogr. 2014;27(9):911–39. 12. Stoichescu-Hogea G, Buleu FN, Christodorescu R, Sosdean R, Tudor A, Ember A, et al. Contribution of Global and Regional Longitudinal Strain for Clinical Assessment of HFpEF in Coronary and Hypertensive Patients. Medicina (Kaunas). 2021;57(12):1372. 13. Soufi Taleb Bendiab N, Meziane-Tani A, Ouabdesselam S, Methia N, Latreche S, Henaoui L, et al. Factors associated with global longitudinal strain decline in hypertensive patients with normal left ventricular ejection fraction. Eur J Prev Cardiol. 2017;24(14):1463–72. 14. Saito M, Khan F, Stoklosa T, Iannaccone A, Negishi K, Marwick TH. Prognostic Implications of LV Strain Risk Score in Asymptomatic Patients With Hypertensive Heart Disease. JACC Cardiovasc Imaging. 2016;9(8):911–21. 15. Menanga A, Nganou-Gnindjio C, Ngo Ngue T, Tiwa Meli D, Sadeu Wafeu G, et al. Alteration of left ventricular global longitudinal function by 2D-strain in a group of Cameroonians heart failure patients with a preserved ejection fraction. Cardiol Trop. 2022;(169):16–21. Available from: https://tropical-cardiology.com (accessed 14 September 2026) 16. Pessinaba S, Afassinou Y, Houssouna D, Kpélafia M, Kaziga W, Togbossi E, et al. Patterns and Predictors of Abnormal Left Ventricular Global Longitudinal Strain Among Black Africans with Hypertension at the Campus University Hospital of Lome: A Cross-sectional Study. Cardiol Cardiovasc Res. 2025;9(3):95–105. 17. Mancia G, Kreutz R, Brunström M, Burnier M, Grassi G, Januszewicz A, et al. 2023 ESH Guidelines for the management of arterial hypertension The Task Force for the management of arterial hypertension of the European Society of Hypertension: Endorsed by the International Society of Hypertension (ISH) and the European Renal Association (ERA). J Hypertens. 2023;41(12):1874–2071. 18. Girerd X, Hanon O, Anagnostopoulos K, Ciupek C, Mourad JJ, Consoli S. Évaluation de l’observance du traitement antihypertenseur par un questionnaire : mise au point et utilisation dans un service spécialisé. Presse Med. 2001;30(21):1044–8. 19. Visseren FLJ, Mach F, Smulders YM, Carballo D, Koskinas KC, Bäck M, et al. 2021 ESC Guidelines on cardiovascular disease prevention in clinical practice. Eur Heart J. 2021;42(34):3227–337. 20. Lang RM, Badano LP, Mor-Avi V, Afilalo J, Armstrong A, Ernande L, et al. Recommendations for Cardiac Chamber Quantification by Echocardiography in Adults: An Update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. J Am Soc Echocardiogr. 2015;28(1):1-39. 21. Nagueh SF, Smiseth OA, Appleton CP, Byrd BF, Dokainish H, Edvardsen T, et al. Recommendations for the Evaluation of Left Ventricular Diastolic Function by Echocardiography: An Update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. J Am Soc Echocardiogr. 2016;29(4):277–314. 22. Yingchoncharoen T, Agarwal S, Popović ZB, Marwick TH. Normal ranges of left ventricular strain: a meta-analysis. J Am Soc Echocardiogr. 2013;26(2):185–91. 23. D'Elia N, Caselli S, Kosmala W, Lancellotti P, Morris D, Muraru D, et al. Normal global longitudinal strain: an individual patient meta-analysis. JACC Cardiovasc Imaging. 2020;13(1 Pt 1):167–9. 24. Voigt JU, Pedrizzetti G, Lysyansky P, Marwick TH, Houle H, Baumann R, et al. Definitions for a common standard for 2D speckle tracking echocardiography: consensus document of the EACVI/ASE/Industry Task Force to standardize deformation imaging. Eur Heart J Cardiovasc Imaging. 2015;16(1):1–11. 25. Murtagh F, Legendre P. Ward’s Hierarchical Agglomerative Clustering Method: Which Algorithms Implement Ward’s Criterion? J Classif. 2014;31(3):274–95. 26. Ghasemi A, Zahediasl S. Normality Tests for Statistical Analysis: A Guide for Non-Statisticians. Int J Endocrinol Metab.2012;10(2):486–9. 27. Box GEP, Tidwell PW. Transformation of the Independent Variables. Technometrics. 1962;4(4):531–50. 28. Zou G. A Modified Poisson Regression Approach to Prospective Studies with Binary Data. Am J Epidemiol. 2004;159(7):702–6. 29. Barros AJD, Hirakata VN. Alternatives for logistic regression in cross-sectional studies: an empirical comparison of models that directly estimate the prevalence ratio. BMC Med Res Methodol. 2003;3:21. 30. Heinze G, Wallisch C, Dunkler D. Variable selection – A review and recommendations for the practicing statistician. Biom J. 2018;60(3):431–49. 31. Uno S, Noma H, Gosho M. Firth‐Type Penalized Methods of the Modified Poisson and Least‐Squares Regression Analyses for Binary Outcomes. Biom J. 2024;66(7):e202400004. 32. Morel JG, Bokossa MC, Neerchal NK. Small Sample Correction for the Variance of GEE Estimators. Biom J. 2003;45(4):395–409. 33. Koo TK, Li MY. A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability Research. J Chiropr Med. 2016;15(2):155–63. 34. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476):307–10. 35. Soufi Taleb Bendiab N. Évaluation de la fonction longitudinale par le 2D strain dans la détection précoce de l’atteinte cardiaque chez l’hypertendu [thesis]. Tlemcen, Algeria: University of Tlemcen; 2018 36. Ferruzzi GJ, Campanile A, Visco V, Loria F, Mone P, Masarone D, et al. Subclinical left ventricular dysfunction assessed by global longitudinal strain correlates with mild cognitive impairment in hypertensive patients. Hypertens Res. 2025;48(5):1768–78. 37. Tadic M, Sala C, Carugo S, Mancia G, Grassi G, Cuspidi C. Myocardial strain and left ventricular geometry: a meta-analysis of echocardiographic studies in systemic hypertension. J Hypertens. 2021;39(11):2297–306. |
|
38. Alpert MA, Omran J, Bostick BP. Effects of Obesity on Cardiovascular Hemodynamics, Cardiac Morphology, and Ventricular Function. Curr Obes Rep. 2016;5(4):424–34. 39. Benyounes N, Lang S, Soulat-Dufour L, Obadia M, Gout O, Chevalier G, et al. Can global longitudinal strain predict reduced left ventricular ejection fraction in daily echocardiographic practice? Arch Cardiovasc Dis. 2015;108(1):50–6. |
|
40. Kuznetsova T, Herbots L, Richart T, D’hooge J, Thijs L, Fagard RH, et al. Left ventricular strain and strain rate in a general population. Eur Heart J. 2008;29(16):2014–23. 41. Kishi S, Armstrong AC, Gidding SS, Colangelo LA, Venkatesh BA, Jacobs DR, et al. Association of Obesity in Early Adulthood and Middle Age With Incipient Left Ventricular Dysfunction and Structural Remodeling. JACC Heart Fail. 2014;2(5):500–8. |

