Effect of alcohol on blood pressure

Sara Tasnim; Chantel Tang; Vijaya M Musini; James M Wright

doi:10.1002/14651858.CD012787.pub2

Efecto del alcohol en la presión arterial

Declaraciones de intereses de los autores

Versión publicada: 01 julio 2020 Historial de versiones

https://doi.org/10.1002/14651858.CD012787.pub2

Contraer todo Desplegar todo

Resumen

disponible en

Antecedentes

Más de 2000 millones de personas en todo el mundo consumen alcohol. Es una sustancia de consumo indebido común y su uso puede provocar más de 200 trastornos, incluida la hipertensión. El alcohol tiene efectos agudos y crónicos sobre la presión arterial. El objetivo de esta revisión fue cuantificar los efectos agudos de diferentes dosis de alcohol a lo largo del tiempo sobre la presión arterial y la frecuencia cardíaca en una población adulta.

Objetivos

Objetivo primario

Determinar los efectos a corto plazo relacionados con la dosis de alcohol versus placebo sobre la presión arterial sistólica y la presión arterial diastólica en adultos sanos e hipertensos mayores de 18 años.

Objetivo secundario

Determinar los efectos a corto plazo relacionados con la dosis de alcohol versus placebo en la frecuencia cardíaca en adultos sanos e hipertensos mayores de 18 años.

Métodos de búsqueda

El especialista en información del Grupo Cochrane de Hipertensión (Cochrane Hypertension Group) buscó ensayos controlados aleatorizados en las siguientes bases de datos hasta marzo 2019: registro especializado del Grupo Cochrane de Hipertensión (Cochrane Hypertension Specialised Register), Registro Cochrane Central de Ensayos Controlados (Cochrane Central Register of Controlled Trials, CENTRAL), (2019, Número 2), MEDLINE (desde 1946), Embase (desde 1974), Plataforma de registros internacionales de ensayos clínicos de la Organización Mundial de la Salud y en ClinicalTrials.gov. También se estableció contacto con los autores de los artículos relevantes con respecto a otros trabajos publicados y no publicados. En las búsquedas no hubo restricciones de idioma.

Criterios de selección

Ensayos controlados aleatorizados (ECA) que compararan los efectos de una dosis única de alcohol versus placebo sobre la presión arterial (PA) o la frecuencia cardíaca (FC) en adultos (≥ 18 años de edad).

Obtención y análisis de los datos

Dos autores de la revisión (ST y CT) extrajeron los datos de forma independiente y evaluaron la calidad de los estudios incluidos. También se contactó con los autores de los ensayos para obtener información faltante o poco clara. La medida de resultado fue la diferencia de medias (DM) del placebo con un intervalo de confianza (IC) del 95%, y se utilizó un modelo de efectos fijos para combinar los tamaños del efecto entre los estudios.

Resultados principales

Se incluyeron 32 ECA con 767 participantes. La mayoría de los participantes del estudio eran hombres (N = 642) y estaban sanos. La edad media de los participantes fue de 33 años y el peso corporal medio fue de 78 kilogramos.

Una dosis baja de alcohol (< 14 g) en el plazo de seis horas (dos ECA, N = 28) no afectó a la PA, pero sí aumentó la FC en 5,1 lpm (IC del 95%: 1,9 a 8,2) (evidencia de certeza moderada).

Una dosis mediana de alcohol (14 a 28 g) en un plazo de seis horas (10 ECA, N = 149) disminuyó la presión arterial sistólica (PAS) en 5,6 mmHg (IC del 95%: ‐8,3 a ‐3,0) y la presión arterial diastólica (PAD) en 4,0 mmHg (IC del 95%: ‐6,0 a ‐2,0) y aumentó la FC en 4,6 lpm (IC del 95%: 3,1 a 6,1) (evidencia de certeza moderada para todos).

Una dosis mediana de alcohol en un plazo de siete a 12 horas (cuatro ECA, N = 54) no afectó a la presión arterial o a la FC.

Una dosis mediana de alcohol > 13 horas después del consumo (cuatro ECA, N = 66) no afectó a la presión arterial o a la FC.

Una dosis alta de alcohol (> 30 g) en un plazo de seis horas (16 ECA, N = 418) disminuyó la PAS en 3,5 mmHg (IC del 95%: ‐6,0 a ‐1,0), disminuyó la PAD en 1,9 mmHg (IC del 95%: 3,9 a 0,04), y aumentó la FC en 5,8 lpm (IC del 95%: 4,0 a 7,5). La certeza de la evidencia fue moderada para la PAS y la FC, y fue baja para la PAD.

Una dosis alta de alcohol en un plazo de siete a 12 horas después del consumo (tres ECA, N = 54) disminuyó la PAS en 3,7 mmHg (IC del 95%: ‐7,0 a ‐0,5) y la PAD en 1,7 mmHg (IC del 95%: ‐4,6 a 1,8) y aumentó la FC en 6,2 lpm (IC del 95%: 3,0 a 9,3). La certeza de la evidencia fue moderada para la PAS y la FC, y baja para el PAD.

Una dosis alta de alcohol ≥ 13 horas después del consumo (cuatro ECA, N = 154) aumentó la PAS en 3,7 mmHg (IC del 95%: 2,3 a 5,1), la PAD en 2,4 mmHg (IC del 95%: 0,2 a 4,5), y la FC en 2,7 lpm (IC del 95%: 0,8 a 4,6) (evidencia de certeza moderada para todos).

Conclusiones de los autores

El alcohol en altas dosis tiene un efecto bifásico en la presión arterial; disminuye la presión hasta 12 horas después del consumo y aumenta la presión arterial a partir de las 13 horas después del consumo. Una dosis alta de alcohol aumenta la FC en todo momento hasta 24 horas. Los hallazgos de esta revisión son relevantes, principalmente para los hombres sanos, ya que sólo se incluyó un pequeño número de mujeres en los ensayos incluidos.

PICO

Population

Intervention

Comparison

Outcome

El uso y la enseñanza del modelo PICO están muy extendidos en el ámbito de la atención sanitaria basada en la evidencia para formular preguntas y estrategias de búsqueda y para caracterizar estudios o metanálisis clínicos. PICO son las siglas en inglés de cuatro posibles componentes de una pregunta de investigación: paciente, población o problema; intervención; comparación; desenlace (outcome).

Para saber más sobre el uso del modelo PICO, puede consultar el Manual Cochrane.

Resumen en términos sencillos

disponible en

El alcohol tiene un efecto bifásico en la tensión arterial y aumenta la frecuencia cardíaca

Pregunta de la revisión

Se revisó la evidencia disponible sobre los efectos a corto plazo de diferentes dosis de bebidas alcohólicas en comparación con las bebidas no alcohólicas sobre la tensión arterial y la frecuencia cardíaca en adultos (≥ 18 años) con tensión normal y alta.

Antecedentes

El consumo excesivo de alcohol se considera una de las causas más habituales del aumento de la presión arterial. Se quiso cuantificar los efectos de una sola dosis de alcohol sobre la presión arterial y la frecuencia cardíaca dentro de las 24 horas posteriores al consumo.

Características de los estudios

Se incluyeron 32 ensayos controlados aleatorizados con 767 participantes publicados hasta marzo de 2019. Aunque estos ensayos incluyeron a adultos de 18 a 96 años de edad con diversos problemas de salud, la mayoría de los participantes en el estudio eran hombres jóvenes y sanos. En la mayoría de los estudios no se informó sobre la fuente de financiación.

Resultados clave

Para dosis bajas de alcohol, se encontró que un vaso de alcohol tenía poco o ningún efecto en la presión arterial y aumentaba la frecuencia cardíaca en las seis horas posteriores a la ingesta.

Existe certeza moderada de que una dosis mediana de alcohol redujo la presión arterial y aumentó la frecuencia cardíaca dentro de las seis horas posteriores al consumo. No se observó ningún cambio significativo en la presión arterial ni en la frecuencia cardíaca después de eso, pero la evidencia fue limitada.

También existe certeza moderada de que una alta dosis de alcohol disminuyó la presión arterial en las seis horas posteriores, y el efecto duró hasta 12 horas. Después de eso, se observó que la presión arterial aumentó. La frecuencia cardíaca aumentó significativamente después del consumo de alcohol y se mantuvo alta en todos los momentos medidos.

Por lo tanto, el alcohol disminuye la tensión arterial inicialmente (hasta 12 horas después de la ingestión) y la aumenta después. El alcohol aumenta constantemente la frecuencia cardíaca en todo momento dentro de las 24 horas posteriores a su consumo.

Authors' conclusions

Implications for practice

The magnitude and direction of the effects of alcohol on blood pressure depend on the time after alcohol consumption. Moderate‐certainty evidence shows that acute consumption of medium to high doses of alcohol decreases blood pressure within the first six hours and for up to 12 hours after alcohol consumption. For times greater than 13 hours, high doses of alcohol consumption increased blood pressure. Low, moderate, and high alcohol consumption increased heart rate within the first six hours. High alcohol consumption also increased heart rate from 7 to 12 hours and after 13 hours. Most of the evidence from this review is relevant to healthy males, as these trials included small numbers of women (126 females compared to 638 males).

Implications for research

This review did not find any eligible RCTs that reported the effects of alcohol on women separately. Because women could be affected differently by alcohol than men, future RCTs in women are needed. If future RCTs include both men and women, it is important that their blood pressure and heart rate readings are reported separately. Although eligible studies included East Asian, Latino, and Caucasian populations, they lacked African, South Asian, and Native Hawaiian/other Pacific Islander representation. Most of the hypertensive participants in the included studies were Japanese, so it is unclear if the difference in blood pressure between alcohol and placebo groups was due to the presence of genetic variants or the presence of hypertension. Large RCTs including both hypertensive and normotensive participants with various ethnic backgrounds are required to understand the effects of alcohol on blood pressure and heart rate based on ethnicity and the presence of hypertension. More RCTs are needed to study the effects of low‐dose alcohol to better delineate the dose‐response effects of alcohol on BP and heart rate. RCTs with measurements more than 24 hours after alcohol consumption are needed to see how long the effect of high‐dose acute alcohol consumption lasts.

Summary of findings

Open in table viewer

Summary of findings 1. Effect of high‐dose alcohol compared to placebo

Outcomes	Participants (RCTs)	Certainty of the evidence (GRADE)	*Mean difference of high‐dose alcohol compared to placebo (95% CI)**
Effect of high‐dose alcohol compared to placebo
Patient or population: adult participants Setting: ambulatory Intervention: high‐dose alcohol (> 30 g) Comparison: placebo
Systolic blood pressure ‐ ≤ 6 hours	418 (16)	⊕⊕⊕⊝ Moderate^a	‐3.5 mmHg [‐6 to ‐0.5]
Systolic blood pressure ‐ 7 to 12 hours	54 (3)	⊕⊕⊕⊝ Moderate^a	‐3.7 mmHg [‐6.9 to ‐0.5]
Systolic blood pressure ‐ ≥ 13 hours	154 (4)	⊕⊕⊕⊝ Moderate^a	3.7 mmHg [2.3 to 5]
Diastolic blood pressure ‐ ≤ 6 hours	350 (14)	⊕⊕⊝⊝ Low^a,b	‐1.9 mmHg [‐3.9 to 0.04]
Diastolic blood pressure ‐ 7 to 12 hours	54 (5)	⊕⊕⊝⊝ Low^a,b	‐1.6 mmHg [‐4.1 to 0.9]
Diastolic blood pressure ‐ ≥ 13 hours	154 (4)	⊕⊕⊕⊝ Moderate^a	2.4 mmHg [0.3 to 4.5]
Heart rate ‐ ≤ 6 hours	495 (17)	⊕⊕⊕⊝ Moderate^a	5.5 bpm [4.3 to 6.7]
Heart rate ‐ 7 to 12 hours	144 (7)	⊕⊕⊕⊝ Moderate^a	6.2 bpm [3 to 9.3]
Heart rate ‐ ≥ 13 hours	244 (8)	⊕⊕⊕⊝ Moderate^a	2.7 bpm [0.8 to 4.6]
* The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aUnclear risk of selection bias and attrition bias in more than one study. ^b95% confidence interval around the best effect estimate includes both negligible effect and appreciable benefit.

Open in table viewer

Summary of findings 2. Effect of medium‐dose alcohol compared to placebo

Outcomes	Participants (RCTs)	Certainty of the evidence (GRADE)	*Mean difference of medium‐dose alcohol compared to placebo (95% CI)**
Effect of medium‐dose alcohol compared to placebo
Patient or population: adult participants Setting: ambulatory Intervention: medium‐dose alcohol (15 to 30 g) Comparison: placebo
Systolic blood pressure ‐ ≤ 6 hours	149 (10)	⊕⊕⊕⊝ Moderate^a	‐5.63 mmHg [‐8.3 to ‐3]
Systolic blood pressure ‐ 7 to 12 hours	54 (4 )	⊕⊕⊝⊝ Low^a,b,c	‐3.2 mmHg [‐8.4 to 2]
Systolic blood pressure ‐ ≥ 13 hours	66 (5)	⊕⊕⊝⊝ Low^a,b	0.6 mmHg [‐3.9 to 5.2]
Diastolic blood pressure ‐ ≤ 6 hours	149 (10)	⊕⊕⊕⊝ Moderate^c	‐4 mmHg [‐6 to ‐2]
Diastolic blood pressure ‐ 7 to 12 hours	54 (4)	⊕⊕⊝⊝ Low^a,b	‐1.2 mmHg [‐4.3 to 1.9]
Diastolic blood pressure ‐ ≥ 13 hours	66 (5)	⊕⊕⊕⊝ Moderate^b	1.8 mmHg [‐0.9 to 4.5]
Heart rate ‐ ≤ 6 hours	181 (12)	⊕⊕⊕⊝ Moderate^c	4.6 bpm [3.1 to 6.1]
Heart rate ‐ 7 to 12 hours	54 (4)	⊕⊕⊝⊝ Low^a,b	1.2 bpm [‐1.9 to 4.3]
Heart rate ‐ > 13 hours	36 (3)	⊕⊕⊝⊝ Low^a,b	1.4 bpm [‐2.1 to 4.9]
* The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aUnclear risk of selection bias in more than one study. ^b95% confidence interval around the effect estimate includes both appreciable benefit and appreciable harm. ^cUnclear risk of selection bias and attrition bias in more than one study.

Open in table viewer

Summary of findings 3. Effect of low‐dose alcohol compared to placebo

Outcomes	Participants (RCTs)	Certainty of the evidence (GRADE)	*Mean difference of low‐dose alcohol compared to placebo (95% CI)**
Effect of low‐dose alcohol compared to placebo
Patient or population: adult participants Setting: ambulatory Intervention: low‐dose alcohol (≥ 14 g) Comparison: placebo
Systolic blood pressure ‐ ≤ 6 hours	28 (2)	⊕⊕⊝⊝ Low^a,b	‐1.9 mmHg [‐8.4 to 5.4]
Diastolic blood pressure ‐ ≤ 6 hours	28 (2 )	⊕⊕⊝⊝ Low^a,b	‐1.5 mmHg [‐6.9 to 4]
Heart rate ‐ ≤ 6 hours	28 (2)	⊕⊕⊕⊝ Moderate^a	5.1 bpm [1.88 higher to 8.24]
* The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aUnclear risk of selection bias. ^b95% confidence interval around the best effect estimate includes both negligible effect and appreciable benefit.

Background

Description of the condition

Hypertension, or elevated blood pressure, is commonly defined as resting systolic blood pressure (SBP) above 140 mmHg or resting diastolic blood pressure (DBP) above 90 mmHg ‐ or both ‐ assuming that the person is not taking any antihypertensive medication (Poulter 2015). It is one of the most common health conditions; hypertension has an increased prevalence with increasing age and affects up to 31% of the world’s adult population (Mills 2016). Sustained hypertension is associated with increased risk of stroke, myocardial infarction, heart failure, renal failure, blindness, and cognitive impairment (Kannel 1972; WHO 2013). In 2015, approximately 10.7 million deaths around the world were estimated to be attributable to hypertension‐related health complications (GBD 2015).

Hypertension can be genetic or may be due to environmental factors such as poor diet, obesity, tobacco use, excessive alcohol consumption, and sedentary lifestyle (Weber 2014; WHO 2013). A population‐based study showed that the incidence of hypertension is higher in African descendants (36%) than in Caucasians (21%) (Willey 2014). Proper management of hypertension can lead to reduction in cardiovascular complications and mortality (Kostis 1997; SHEP 1991; Staessen 1999). Although pharmacological interventions can effectively reduce blood pressure, multiple studies have shown that a healthy lifestyle alone without any pharmacological interventions can greatly reduce the prevalence of hypertension (Appel 2003; Guitteau 2006). According to US guidelines for prevention, detection, evaluation, and management of high blood pressure, adults are advised to reduce weight and sodium intake, increase physical activity and potassium intake, cease smoking, and moderate alcohol intake to manage hypertension non‐pharmacologically (Reboussin 2018).

Description of the intervention

Alcohol has been a part of almost every human culture for a very long time (McGovern 2009). According to the World Health Organization (WHO), around 2.3 billion people globally drink alcohol, and most of them are from the European region. On average, drinkers consume 32.8 grams of pure alcohol per day, and beer (34.3%) is the most consumed alcoholic beverage (WHO 2018). In the United States, 14 grams of pure alcohol is considered as one standard drink or one unit, and the maximum daily limit for men and women is four and three drinks, respectively (NIAAA 2017). Exceeding this limit increases the risk of cardiovascular, hepatic, and nervous system disorders (Bellentani 1997; Fuchs 2001; Gao 2011; Lieber 1998; McCullough 2011; Nutt 1999; Welch 2011). Also, multiple studies have found associations between consumption of alcoholic beverages and specific cancers (Kushi 2012; Seitz 2007). Abuse of alcohol resulted in approximately 3 million deaths worldwide and 132.6 million disability‐adjusted life years (DALYs) in 2016 (WHO 2018).

Alcohol is water‐soluble and can cross biological membranes by passive diffusion. It reaches equilibrium quickly if the body water content is higher. The presence of food in the stomach slows down alcohol absorption by retarding gastric emptying. Hence, it is recommended to not drink alcoholic beverages on an empty stomach. Alcohol is predominantly metabolised by the alcohol dehydrogenase (ADH) enzyme system and to a lesser extent by cytochrome P‐450 2E1 in the liver. Alcohol is first metabolised to acetaldehyde and then is oxidised into acetyl coenzyme A (CoA) by aldehyde dehydrogenase (ALDH) (Cederbaum 2012). Although alcohol metabolism by the liver is well characterised, its metabolism in other parts of the body is not well defined. The enzyme catalase was found to be responsible for metabolising alcohol in the brain to produce acetaldehyde (Heit 2013). Acetaldehyde is highly reactive and has been associated with a wide range of physiological adverse effects (Zimatkin 2006).

Despite the potential negative effects of alcohol consumption, systematic reviews based on cohort studies show that light to moderate consumption of alcohol has a cardioprotective effect and may decrease mortality in adult men and women (Briasoulis 2012; Di Castelnuovo 2006; Plunk 2014; Taylor 2009).

How the intervention might work

The molecular mechanisms through which alcohol raises blood pressure are unclear. Alcohol can affect blood pressure through a variety of possible mechanisms. Previous research suggests that acute alcohol consumption affects the renin–angiotensin–aldosterone system (RAAS) by increasing plasma renin activity (Puddey 1985). The RAAS is responsible for maintaining the balance of fluid and electrolytes. An increase in plasma renin results in increased production of angiotensin I (AI), which is converted to angiotensin II (AII) by angiotensin‐converting enzyme (ACE). The hormone AII is a potent vasoconstrictor that stimulates aldosterone and vasopressin secretion from the adrenal gland, promoting sodium and water retention (Schrier 1999). As a result, peripheral resistance and blood volume are increased, leading to elevated arterial blood.

Several clinical trials in humans and studies conducted in animal models have reported stimulation of the sympathetic nervous system and increased noradrenaline after consumption of alcohol (Barden 2013; Grassi 1989; Randin 1995; Russ 1991; Zhang 1989). When noradrenaline stimulates the adrenergic receptors located in the heart muscles, heart rate and blood pressure are increased.

Alcohol has been reported to diminish baroreceptor sensitivity, which is a key factor in regulating blood pressure (Abdel‐Rahman 1985; Rupp 1996). Baroreceptors or stretch receptors are mechanoreceptors located on the arch of the aorta and the carotid sinus. They can detect changes in blood pressure and can maintain blood pressure by controlling heart rate, contractility, and peripheral resistance. Acute administration of alcohol stimulates the release of histamine and endorphin, which interferes with baroreflex sensitivity (Carretta 1988).

Another possible mechanism is the increase in plasma cortisol levels following heavy alcohol consumption (Jenkins 1968). Several studies have suggested a role for cortisol in alcohol‐induced hypertension (Husain 2014; Potter 1986). Cortisol is a type of steroid hormone, and the presence of excess cortisol has been associated with elevated blood pressure in normotensive individuals (Whitworth 1984; Whitworth 2005).

Alcohol can affect drinkers differently based on their age, sex, ethnicity, family history, and liver condition (Cederbaum 2012; Chen 1999; Gentry 2000; Thomasson 1995). Previous studies reported that women are affected more than men after drinking the same amount of alcohol because of their lower body weight and higher body fat. The blood alcohol concentration (BAC) rises faster in women because they have a smaller volume of distribution (Kwo 1998). In contrast, women eliminate alcohol from the body a little faster than men (Thomasson 2000). Different genetic variants of ADH and ALDH enzymes have been found to show strikingly different rates of alcohol metabolism among different races (Chen 1999; Peng 2014; Agarwal 1981).

Why it is important to do this review

Several systematic reviews based on cohort studies have concluded that alcohol intake has a considerable effect on blood pressure and on risk of hypertension (Chen 2008; Worm 2013). It has also been shown that heavy alcohol consumption causes hypertension and leads to left ventricular dysfunction and dilated cardiomyopathy. On the other hand, abundant epidemiological and clinical evidence shows that light to moderate drinking is associated with reduced risk of coronary heart disease (CHD), incidence of stroke, and total mortality among middle‐aged and elderly men and women (Abramson 2001;Briasoulis 2012; Di Castelnuovo 2006; Djoussé 2007;Jaubert 2014; Plunk 2014; Taylor 2009).

All these conclusions are based on findings of observational studies. Several RCTs have reported the magnitude of effect of alcohol on blood pressure, but because those trials are small, their findings are not sufficient to justify a strong conclusion. In 2005, McFadden and colleagues conducted a systematic review of RCTs, which investigated the haemodynamic effects of daily consumption of alcohol (McFadden 2005). Based on nine RCTs in which participants consumed alcohol repeatedly over days, these review authors reported that alcohol increases SBP by 2.7 mmHg and DBP by 1.4 mmHg. However, they excluded studies for which the duration of BP observation was less than 24 hours and articles published in non‐English languages. We believe that inclusion of those studies will provide useful information about the dose‐related magnitude and time‐course effect of alcohol on blood pressure in people with both normal and elevated blood pressure.

Objectives

Primary objectives

To determine short‐term dose‐related effects of alcohol versus placebo on systolic blood pressure and diastolic blood pressure in healthy and hypertensive adults over 18 years of age.

Secondary objective

To determine short‐term dose‐related effects of alcohol versus placebo on heart rate in healthy and hypertensive adults over 18 years of age.

Methods

Criteria for considering studies for this review

Types of studies

All randomised controlled trials (RCTs) that compared alcohol to placebo or similar tasting non‐alcoholic beverages were included in this systematic review.

Types of participants

We included adult (≥ 18) participants of both sexes without any restriction on their health condition.

Types of interventions

We included any type of alcoholic beverage as the intervention arm. The dose of alcohol had to be reported by study authors for inclusion in the systematic review. Because there are no published standards for differentiating between low and medium doses of alcohol, we chose the alcohol content in one standard drink as the threshold between low dose and medium dose. Because the alcohol content in one standard drink varies among different countries (ranging from 8 g to 14 g), we chose the Canadian standard for an alcoholic beverage, which is 14 g of pure alcohol (CCSA). Accordingly, we considered up to 14 g of alcohol as a low dose of alcohol. To differentiate between medium and high doses, the Canadian Centre on Substance Use and Addiction (CCSA) identifies less than 30 g of alcohol for men and less than 20 g of alcohol for women as the threshold for low risk of alcohol intake (CCSA). Thus, in our review, we used up to 30 g alcohol intake for men and up to 20 g alcohol intake for women as a moderate dose, and above this limit as a high dose. In studies where sex‐specific results were not provided, we categorised dose based on the dominating sex in terms of study participation. In conclusion, we categorised doses of alcohol as follows.

Low dose (≤ 14 g of alcohol/≤ 1 standard drink).
Medium dose (> 14 g and ≤ 30 g of alcohol for men and > 14 g and ≤ 20 g of alcohol for women).
High dose (> 30 g of alcohol for men and > 20 g of alcohol for women).

Types of outcome measures

Primary outcomes

Change in resting seated systolic and diastolic blood pressures at three different time periods after alcohol intake: early (up to six hours); intermediate (7 to 12 hours); and late (≥ 13 hours)

Secondary outcomes

Change in resting heart rate at the same time periods as blood pressure outcomes above

Additional outcomes

Change in resting mean arterial pressure (MAP) at the same time periods as blood pressure outcomes above

Search methods for identification of studies

Electronic searches

The Cochrane Hypertension Information Specialist searched the following databases without language, publication year, or publication status restrictions.

Cochrane Hypertension Specialised Register via the Cochrane Register of Studies (CRS‐Web) (searched 14 March 2019).
Cochrane Central Register of Controlled Trials (CENTRAL; 2019, Issue 2), in the Cochrane Library, via the Cochrane Register of Studies (CRS‐Web) (searched 14 March 2019).
MEDLINE Ovid (from 1946 onwards), MEDLINE Ovid Epub Ahead of Print, and MEDLINE Ovid In‐Process & Other Non‐Indexed Citations (searched 14 March 2019).
Embase Ovid (from 1974 onwards) (searched 14 March 2019).
ClinicalTrials.gov (www.clinicaltrials.gov) (searched 14 March 2019).
World Health Organization International Clinical Trials Registry Platform (www.who.it.trialsearch) (searched 14 March 2019).

The Information Specialist modelled subject strategies for databases on the search strategy designed for MEDLINE. When appropriate, these were combined with subject strategy adaptations of the highly sensitive search strategy designed by Cochrane for identifying RCTs (as described in the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0, Box 6.4.c. (Higgins 2011)). We present search strategies for major databases in Appendix 1.

Searching other resources

The Cochrane Hypertension Information Specialist searched the Hypertension Specialised Register segment (which includes searches of MEDLINE, Embase, and Epistemonikos for systematic reviews) to retrieve existing reviews relevant to this systematic review, so that we could scan their reference lists for additional trials. The Specialised Register also includes searches of CAB Abstracts & Global Health, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), ProQuest Dissertations & Theses, and Web of Science for controlled trials
We checked the bibliographies of included studies and any relevant systematic reviews identified for further references to relevant trials
When necessary, we contacted authors of key papers and abstracts to request additional information about their trials

Data collection and analysis

Selection of studies

We (ST and CT) independently screened the citations found through the database search using Covidence software (Covidence). We excluded articles if the citation seemed completely irrelevant or was identified as a review or observational study after the title and abstract were read. For remaining studies, we (ST and CT) retrieved full‐text articles for further assessment. Any disagreements regarding inclusion or exclusion of studies were resolved by discussion between review authors. The reason for exclusion was documented for each citation at the full‐text level. We also checked the list of references in the included studies and articles that cited the included studies in Google Scholar to identify relevant articles. We reported the flow of citation in Figure 1.

Figure 1

Study flow diagram.

Data extraction and management

Two review authors (ST and CT) performed data extraction independently using a standard data collection form, followed by a cross‐check. In cases of disagreement, the third review authors (JMW) became involved to resolve the disagreement. When necessary, we contacted the authors of studies for information about unclear study design. We recorded study design, type of masking, randomisation and allocation concealment methods, details of intervention and comparator, duration of intervention, baseline characteristics of participants, whether food was consumed before or during the intervention, numbers of participants included in the final analysis, outcomes and results, method and position of BP measurement, declaration of conflict of interest, funding source, and protocol registration number. All extracted data were entered and double‐checked in RevMan 5.3 software (Review Manager (RevMan)).

Assessment of risk of bias in included studies

We (ST and CT) assessed the risk of bias of included studies independently using the Cochrane risk of bias tool (version 1) according to Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions for the following domains (Higgins 2011).

Sequence generation (selection bias).
Allocation sequence concealment (selection bias).
Blinding of participants (performance bias).
Blinding of outcome assessors (detection bias).
Incomplete outcome data (attrition bias).
Selective outcome reporting (reporting bias).
Other potential sources of bias (i.e. conflict of interest, funding source, registration of the study protocol).

We assessed selective reporting bias for each of the outcomes separately. For the other domains, we grouped outcomes together and provided only one judgement. We contacted study authors for missing or unclear information required for the risk of bias assessment and then reassessed the domains once the information was available.

Measures of treatment effect

All outcomes of interest in the review (BP and HR) produced continuous data. We calculated and reported mean difference (MD), with corresponding 95% confidence interval (95% CI).

Unit of analysis issues

Most of the studies included in the review had a cross‐over design. The carry‐over effect in a cross‐over trial can confound the effects of subsequent treatment. We recorded the washout period of each included study reported by study authors to decide if there was risk of a carry‐over effect. If it was appropriate to combine cross‐over trials with other trials, we used the recommended generic inverse variance approach of meta‐analysis. We tested the effect of cross‐over trials through sensitivity analysis by excluding them from the meta‐analysis to check if the effect estimate changed significantly.

For multi‐arm trials, if a study reported more than one intervention arm, we identified the relevant intervention arm and included that in the review. If studies reported more than one placebo group, we combined them into a single group when appropriate, using the formulae for combining groups reported in Chapter 7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We followed the same formulae for combining groups if a study reported two different types of alcoholic beverages containing the same amount of alcohol.

Dealing with missing data

General missing data

We contacted the study authors for missing or unclear information relevant to the review using contact information provided in their respective articles. If the dose of a study was not reported in the article and the study author did not respond to our request, we excluded that study.

Missing statistics

If a standard error (SE) was given instead of a standard deviation (SD), we used the formula SD = SE × square root of n (number of participants) to calculate the SD.

We also calculated SD if 95% CI, P value, or t value was reported in the included studies, according to Chapter 7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). If we were not able to get SD from the study authors or calculate SD from the values mentioned above, we imputed SD using the following hierarchy (listed from highest to lowest) (Musini 2014).

SD of change in blood pressure measured in a different position (e.g. lying down) than that of the blood pressure data used.
SD of blood pressure at the end of treatment.
SD of blood pressure at the end of treatment measured in a different position (e.g. lying down) than that of the blood pressure data used.
SD of blood pressure at baseline (unless this measure was used as an entry criterion).
Mean SD of change in blood pressure from other studies that studied the effects of alcohol.

Assessment of heterogeneity

We conducted a standard Chi² test through Review Manager Software 5.3 to test for heterogeneity (Review Manager (RevMan)). A P value of 0.1 or less was considered to show statistically significant heterogeneity. The I² statistic was used to interpret the level of heterogeneity (Higgins 2011). For the purposes of this review, if I² was greater than 50%, it was considered to show a substantial level of heterogeneity. Furthermore, we visually inspected the forest plot to check whether there were any non‐overlapping confidence intervals indicating heterogeneity. Last, we attempted to explore the reason for heterogeneity by looking for clinical and methodological differences between trials.

Assessment of reporting biases

We used funnel plots if there were more than 10 studies that contributed to a meta‐analysis to detect the extent of risk of reporting bias based on symmetry of the plot (Higgins 2011). We visually inspected the funnel plots. If the residual scatter plot resembles a symmetrical inverted funnel, we considered publication bias to be unlikely. Positive results are more likely to be published than negative results, which leads to potential publication bias. However, publication bias does not necessarily lead to asymmetry in funnel plots. Asymmetry in the funnel plot can also be due to inflated effects in smaller studies resulting from poor study design, heterogeneity, sampling variation, or chance. Therefore, we performed sensitivity analyses and searched for unpublished studies as outlined in Chapter 10 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011)

Data synthesis

We used Cochrane review manager software for all data analyses (Review Manager (RevMan)). We conducted meta‐analysis for the three dose groups (low dose, medium dose, and high dose of alcohol) separately. We considered statistical, clinical, and methodological heterogeneity between study populations and proceeded with the meta‐analysis if only we considered interventions, comparisons, and outcome measures similar enough to pool. When trials compared more than one dose of alcohol, we handled each comparison separately. Because all of our outcomes of interest provided continuous data, we used the inverse variance approach and a fixed‐effect model to combine effect sizes across studies.

Subgroup analysis and investigation of heterogeneity

We planned on performing subgroup analysis based on the following.

Normotensive participants (defined as SBP < 140 mmHg and DBP < 90 mmHg) versus hypertensive participants (SBP ≥ 140 mmHg or DBP ≥ 90 mmHg).
Sex of participants.

It is recommended that there should be at least 10 studies reporting each of the subgroups in question (Deeks 2011). Among the 34 included studies, only four studies included hypertensive participants. So, it was not possible to conduct a subgroup analysis based on blood pressure. For the planned subgroup analysis based on sex, no study reported male and female participant data separately.

Sensitivity analysis

We performed the following sensitivity analyses.

We checked if blinding of participants and outcome assessors affected the effect estimate of BP and HR (blinded studies versus unblinded studies).
We checked the difference between effect estimates of outcomes given by the fixed‐effect model and the random‐effects model by conducting sensitivity analysis. We did this only when heterogeneity was substantial.

Summary of findings and assessment of the certainty of evidence

We used the GRADE approach (Grading of Recommendations, Assessment, Development and Evaluation) to assess the certainty of a body of evidence as high, moderate, low, or very low and provided review authors' comments to support our judgements as outlined in Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions ((Guyatt 2011 Higgins 2011). We included the outcomes SBP, DBP, and HR for each comparison. Both review authors (ST and CT) rated the certainty of evidence independently by examining risk of bias, indirectness, inconsistency, imprecision, and publication bias.

To assess risk of bias across studies, we rated the evidence as having no limitations, serious limitations, or very serious limitations while taking into account the extent that each trial contributes towards the magnitude of effect (weight) as based on its study sample size and mean difference.

We used GRADEpro software to construct a 'Summary of findings' table to compare outcomes including change in SBP and DBP and HR (GRADEpro 2014). In addition, we included illustrative risks to present findings for the most important outcome (change in systolic blood pressure).

Summary of findings and assessment of the certainty of the evidence

We used the GRADE approach (Grading of Recommendations, Assessment, Development and Evaluation) (Guyatt 2011) to assess the certainty of the body evidence as high, moderate, low or very low and provide review authors' comments to support our judgements as outlined in the Cochrane Handbook for Systematic Reviews of Interventions chapter 12 (Higgins 2011). We included the outcomes SBP, DBP and HR for each comparison. Both reviewers (ST and CT) rated the certainty of evidence independently by examining risk of bias, indirectness, inconsistency, imprecision, and publication bias.

To assess the risk of bias across studies, we rated the evidence as having no limitations, serious limitations, or very serious limitations while taking into account the extent that each trial contributes towards the magnitude of effect (weight) as based on their study sample size and mean difference.

We used GRADEpro software (GRADEpro 2014) to construct a 'Summary of findings' table to compare outcomes including change in SBP and DBP and HR. In addition, we also included illustrative risks to present findings for the most important outcome (change in systolic blood pressure)

Results

Description of studies

Results of the search

The search was conducted up to March 2019 and resulted in 6869 citations. After de‐duplication and screening of titles and abstracts, we were left with 482 citations for further assessment. We retrieved full‐text articles for those citations and included 32 studies (Figure 1).

Included studies

Refer to Characteristics of included studies and Table 1 for further details regarding these studies.

Open in table viewer

Table 1. Baseline characteristics

Study ID	Randomised participants, N	Mean age (range)	Mean body weight, kg	Health condition	Reported dose of alcohol	Duration of intervention	Baseline SBP (SD)	Baseline DBP (SD)	Baseline HR (SD)
Agewall 2000	12	31 (younger than 40 years old)	Not reported	Healthy, normotensive non‐smokers	31.25 g	10 minutes	121 (6)	79 (4)	61 (7)
Barden 2013	24	56 (20 to 65)	Not reported	Healthy	41 g	30 minutes	115 (11)	72 (6)	62 (7)
Bau 2005	100	20.7 (18 to 25)	Not reported	Healthy non‐smokers	60 g	30 minutes	114.2	64.8	72.43 (10.9)
Bau 2011	70	20.7 (18 to 25)	Not reported	Healthy	60 g	30 minutes	Not reported	Not reported	75 (10)
Buckman 2015	72	21.5	Not reported	Healthy	0.90 mL/kg for men 0.78 mL/kg for women	15 minutes	116.9 (13.5)	Not reported	66.9 (9.9)
Chen 1986	20	19 to 32	Not reported	Healthy, normotensive non‐smokers	Target to achieve blood level of 0.05%	Not reported (mentioned that fairly fast rate)	118 (12.88)	62.25 (5.1)	63.13 (7.1)
Cheyne 2004	17	35 (21 to 46)	Not reported	Type 1 diabetes	0.35 mg/kg BW	Not reported	116.2 (18.7)	66.8 (8.2)	70 (12)
Dai 2002	40	19 to 25 years	81.35	Healthy	0.5 g/kg	5 minutes	114.1	72.6	63.5
Dumont 2010	14	22.1 (18 to 29)	Not reported	Regular user of MDMA, otherwise healthy	Target blood alcohol level 0.6%, equivalent of 2 to 3 alcoholic beverages.	3 hours to maintain target BAC	Not reported	Not reported	66
Fantin 2016	18	34.2 (25 to 53)	70.2 (53 to 85.6)	Healthy	30 g	10 hours	110.3 (12)	80 (8)	75.5 (11.5)
Fazio 2004	10	22 (20 to 25)	Not reported	Healthy	0.3 g/kg	5 minutes	Not reported	Not reported	68
Foppa 2002	13	55 (43 to 65)	Not reported	Hypertensive and centrally obese	23 g	15 minutes	130	83	72 (8.72)
Hering 2011	24	44	Not reported	13 hypertensive and 11 normotensive	1 g/kg	20 minutes	Hypertensive: 150 (21) Normotensive: 136 (13.2)	Hypertensive: 91 (14.4) Normotensive: 76 (10)	Hypertensive: 72 (7.2) Normotensive: 70 (10)
Karatzi 2005	15	52.4	Not reported	Coronary artery disease	30 g	Not reported	109.8 (9.2)	80.7 (10.8)	67.1 (13.1)
Karatzi 2013	16	28.5	77.5	Healthy non‐smokers	20 g	15 minutes	115.4 (6.2)	68.5 (5.4)	60 (8.1)
Kawano 1992	13	55.2 (22 to 70)	65.2	Mild to moderate essential hypertension	51 g	Not reported	159 (18.8)	91.3 (12)	61.5
Kawano 2000	10	54 (32 to 67)	70 (60 to 78)	Mild essential hypertension	55.3 g (1 mL/kg BW)	60 minutes	147	91	65
Koenig 1997	15	20 to 35	76	Healthy	10 mL/kg BW	30 minutes	127 (11)	80 (9.5)	Not reported
Kojima 1993	21	56.5 (33 to 73)	Not reported	Essential hypertension	1 mL/kg BW	30 minutes	146 (18.33)	89 (9.2)	59 (9.2)
Mahmud 2002	8	21 to 40	70	Healthy, normotensive non‐smokers	56 g (0.8 g/kg of BW)	10 minutes	93.3 (10)	67 (8)	Not reported
Maufrais 2017	24	23.3	62.9	Healthy	0.4 g/kg	5 minutes	Not reported	Not reported	69 (9.8)
Maule 1993	10	31 (22 to 51)	68.1 (50 to 81)	Healthy	34 g (0.5 g/kg BW)	10 minutes	122	70	62 (6.3)
Narkiewicz 2000	19	26	Not reported	Healthy	1 g/kg BW	30 minutes	111 (11.2)	61 (7.5)	57 (7.5)
Nishiwaki 2017	11	21.1 (20 to 22)	62.6	Healthy non‐smokers	11 g and 19.25 g	5 minutes	123 (6.63)	71 (6.63)	59 (9.94)
Potter 1986	16	22 (20 to 30)	77	Healthy, normotensive	0.75 g/kg	15 minutes	122.5 (11)	72.5 (9)	63.2 (8)
Rosito 1999	40	22.2 (19 to 30)	Not reported	Healthy	60 g	1 hour	124.2 (10.7)	76.2 (9.4)	70.5 (12.6)
Rossinen 1997	20	39 to 68	Not reported	Coronary artery disease and myocardial ischaemia Patients were taking usual medicine	1.25 g/kg	1 hour 30 minutes	132 (16)	Not reported	69.5
Stott 1987	10	18 to 31	56 to 101	Healthy	1.3 g/kg	1 hour	115.5	67	70.5
Stott 1991	8	81 (70 to 96)	68.4	Normotensive	0.5 g/kg	15 minutes	130 (18.3)	77.5 (15.5)	57 (7.5)
Van De Borne 1997	16	26	Not reported	Healthy	1 g/kg	30 minutes	Not reported	Not reported	59 (8)
Williams 2004	13	59 (48 to 70)	86	Coronary artery disease	0.52 g/kg	20 minutes	135 (11)	82 (7)	55 (11)
Zeichner 1985	48	20.9 (19 to 23)	Not reported	Healthy	1 g/kg	20 minutes	115.5 (10.2)	70.4 (9.1)	69.4 (12.10)

BAC: blood alcohol concentration.
BW: body weight.
DBP: diastolic blood pressure.
HR: heart rate.
SBP: systolic blood pressure.
SD: standard deviation.

Of the 32 included RCTs involving 767 participants, 26 trials used a cross‐over design and six used a parallel‐group design. Three studies were single‐blind, 12 were double‐blind, and 17 were open‐label studies. Most study participants were male (N = 642). The age of participants ranged from 18 to 96 years, and mean participant age was 33 years. Stott 1991 included relatively old participants (mean age 81, range 70 to 96 years) compared to the other studies. Only 14 out of 34 studies reported the mean body weight of participants. The mean body weight from those 14 studies was 78 kg. Most studies included healthy participants with normal blood pressure. Some studies included small numbers of participants with essential hypertension (10%) (Foppa 2002 Hering 2011; Kawano 1992; Kawano 2000; Kojima 1993), type 1 diabetes (2.2%) (Cheyne 2004), coronary heart disease (6.3%) (Karatzi 2005 Rossinen 1997 Williams 2004) and regular user of methylenedioxymethamphetamine (MDMA) (2%) (Dumont 2010).

Different types of alcoholic beverages including red wine, white wine, beer, and vodka were used among 32 studies. Out of 32 studies, 23 used non‐alcoholic beverages including juice as placebo (Bau 2005; Bau 2011 Buckman 2015; Cheyne 2004; Dai 2002; Dumont 2010; Fazio 2004; Foppa 2002; Hering 2011; Kawano 1992; Kawano 2000; Koenig 1997; Kojima 1993; Maufrais 2017; Maule 1993; Narkiewicz 2000; Rosito 1999; Rossinen 1997; Stott 1987; Stott 1991; Van De Borne 1997; Williams 2004; Zeichner 1985); seven studies used de‐alcoholised wine as placebo (Agewall 2000; Barden 2013; Karatzi 2005; Karatzi 2013; Mahmud 2002; Nishiwaki 2017; Potter 1986); and two studies did not mention the type of control they used for alcohol (Chen 1986; Fantin 2016). The dose of alcohol ranged between 0.35 mg/kg and 1.3 g/kg, and alcohol was consumed over five minutes and over one hour and 30 minutes. It is important to note that the dose of alcohol was comparatively higher (≥ 60 g or ≥ 1 g/kg) in nine studies (Bau 2005; Buckman 2015; Hering 2011; Narkiewicz 2000; Rosito 1999; Rossinen 1997; Stott 1987; Van De Borne 1997; Zeichner 1985).

Excluded studies

We excluded 450 trials after reviewing the full‐text articles, and we recorded the reasons for exclusion (see table Characteristics of excluded studies table).

Risk of bias in included studies

Refer to Figure 2 and Figure 3 for the overall 'Risk of bias' assessment.

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figure 3

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

We (ST and CT) independently assessed risk of bias by following the methods described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We assessed the risk of bias based on 11 domains: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias) for systolic blood pressure (SBP), selective reporting (reporting bias) for diastolic blood pressure (DBP), selective reporting (reporting bias) for mean arterial blood pressure (MAP), selective reporting (reporting bias) for heart rate (HR), other bias (conflict of interest, industry sponsorship), and other bias (was the study registered in clinical trials.gov and was the protocol available). We classified each domain as being at low, high, or uncertain risk of bias.

In the case of disagreement, a third party (JMW) was involved to discuss and resolve the disagreement. In the case of uncertain information regarding the method of RCT, we contacted study authors via email to request clarification. Refer to Table 2 for further details regarding reasons and responses.

Open in table viewer

Table 2. Contact with corresponding authors

Study ID	Reason for contact	Contacted? (Yes/No)	Response? (Yes/No)
Agewall 2000	Method of allocation concealment used in RCT was not mentioned	No Comment ‐ contact information cannot be found	NA
Bau 2005	Method of allocation concealment used in RCT was not mentioned	Yes	No
Bau 2011	Method of allocation concealment used in RCT was not mentioned	Yes	No
Buckman 2015	Method of allocation concealment used in RCT was not mentioned	Yes	Yes
Chen 1986	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Dai 2002	Method of allocation concealment used in RCT was not mentioned	Yes	No
Dumont 2010	Methods of randomisation and allocation concealment, blinding of participants and personnel, and blinding of outcome assessment used in RCT were not mentioned	Yes	No
Fantin 2016	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Fazio 2004	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Foppa 2002	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	Yes
Hering 2011	Methods of allocation concealment used in RCT was not mentioned	Yes	Yes
Karatzi 2005	Methods of randomisation and allocation concealment, blinding of participants and personnel, and blinding of outcome assessment used in RCT were not mentioned	Yes	No
Karatzi 2013	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Kawano 1992	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes Comment ‐ contact information cannot be found in the study. However, we used contact information provided in Kawano 2000	No
Kawano 2000	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Kojima 1993	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Mahmud 2002	Methods of randomisation and allocation concealment, blinding of participants and personnel, and blinding of outcome assessment used in RCT were not mentioned	Yes	No
Maufrais 2017	Method of allocation concealment used in RCT was not mentioned	Yes	No
Maule 1993	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Narkiewicz 2000	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Nishiwaki 2017	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Potter 1986	Methods of randomisation and allocation concealment and blinding of outcome assessor used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Rossinen 1997	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Rosito 1999	Methods of randomisation and allocation concealment and blinding of participants and personnel used in RCT were not mentioned	Yes	Yes
Stott 1987	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Stott 1991	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Van De Borne 1997	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Williams 2004	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Zeichner 1985	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA

NA: not applicable.
RCT: randomised controlled trial.

Allocation

We (ST and CT) assessed selection bias based on two categories: random sequence generation and allocation concealment.

Random sequence generation

For random sequence generation, we classified 22 included studies as having uncertain risk of bias (Agewall 2000; Chen 1986; Dumont 2010; Fantin 2016; Fazio 2004; Karatzi 2005; Karatzi 2013; Kawano 1992; Kawano 2000; Koenig 1997; Kojima 1993; Mahmud 2002; Maule 1993; Narkiewicz 2000; Nishiwaki 2017; Potter 1986; Rossinen 1997; Stott 1987; Stott 1991; Van De Borne 1997; Williams 2004; Zeichner 1985). Even though these studies reported that participants were randomised to receive alcohol or placebo, the method of randomisation was not mentioned. Although three studies did not report the method of randomisation (Barden 2013; Buckman 2015; Dai 2002), their reported baseline characteristics were well matched. We classified them as having uncertain risk of bias. The remaining seven studies reported the method of randomisation used, hence we classified them as having low risk of bias. Random seed generation was used in Bau 2005 and Bau 2011 computer‐generated random selection of 4 × 4 Latin squares was used in Cheyne 2004 a third person unaware of research objectives or protocol prepared sealed randomised envelops in blocks of eight in Foppa 2002; a randomised computer‐generated number table was used in Hering 2011; a random sequence generator was used in Maufrais 2017; and a random number allocator was used in Rosito 1999. It is important to note that information regarding to the method of randomisation used in Foppa 2002 and Rosito 1999 was provided by the study author via email. Refer to Table 2 for further details.

For allocation concealment, we classified 28 included studies as having uncertain risk of bias because the method of allocation concealment was not reported (Agewall 2000; Bau 2005; Bau 2011; Buckman 2015; Chen 1986; Dai 2002; Dumont 2010; Fantin 2016; Fazio 2004; Hering 2011; Karatzi 2005; Karatzi 2013; Kawano 1992; Kawano 2000; Koenig 1997; Kojima 1993; Mahmud 2002; Maufrais 2017; Maule 1993; Narkiewicz 2000; Nishiwaki 2017; Potter 1986; Rossinen 1997; Stott 1987; Stott 1991; Van De Borne 1997; Williams 2004; Zeichner 1985). We classified the remaining four studies as having low risk of bias. In Barden 2013, treatment allocation was performed by a statistician who was not involved in the trial. Opaque sealed randomised envelopes were used in Cheyne 2004 and Foppa 2002, and random number allocator was used in Rosito 1999. It is important to note that information regarding the method of allocation concealment used in Foppa 2002 and Rosito 1999 was provided by the study author via email. We also contacted Hering 2011, but the study author did not explicitly mention in the email the method of allocation concealment used. Thus, we classified this study as having uncertain risk of bias. Refer to Table 2 for further details.

Blinding

In the case of performance bias, we classified six studies as having low risk of bias, 19 studies as having high risk of bias, and seven studies as having unclear risk of bias.

We classified six studies as having low risk of performance bias (Dai 2002; Narkiewicz 2000; Nishiwaki 2017; Potter 1986; Rosito 1999; Van De Borne 1997). Nishiwaki 2017 was a single‐blinded study. In this study, all test drinks were poured into paper cups to achieve blinding of participants. We contacted the author of Rosito 1999 to request additional information regarding the method of blinding used. The study author explained the blinding method in detail in an email, so we classified this study as having low risk of bias.

We classified 19 studies as having high risk of performance bias. Of 19 studies, 17 were open‐label (Barden 2013; Buckman 2015; Chen 1986; Fantin 2016; Fazio 2004; Foppa 2002; Hering 2011; Kawano 1992; Kawano 2000; Koenig 1997; Kojima 1993; Maufrais 2017; Rossinen 1997; Stott 1987; Stott 1991; Williams 2004; Zeichner 1985), hence these studies were not blinded for participants nor for personnel. It is important to note that 2 out of 19 studies were single‐blinded (Agewall 2000; Karatzi 2013). Personnel were blinded instead of participants in Karatzi 2013, and neither personnel nor participants were blinded in Agewall 2000, so we assessed these studies as having high risk of bias.

We classified seven studies as having unclear risk of performance bias (Bau 2005; Bau 2011; Cheyne 2004; Dumont 2010; Karatzi 2005; Mahmud 2002; Maule 1993). Bau 2005 and Bau 2011 mentioned only that investigators and volunteers were blinded to the content of the drink but did not mention the method of blinding used in these studies. Karatzi 2005 mentioned the method of blinding of participants, but it is not clear whether involved personnel were blinded as well. The method of blinding of participants and personnel was not mentioned in Dumont 2010, Mahmud 2002, and Maule 1993. In Cheyne 2004, participants were blinded to the content of the drink, but some reported that they were able to detect the alcohol by taste at the end of the study. Hence, we classified this study as having high risk of bias.

In the case of detection bias, we classified nine studies as having low risk of performance bias (Agewall 2000; Bau 2005; Bau 2011; Cheyne 2004; Dai 2002; Karatzi 2013; Narkiewicz 2000; Rosito 1999; Van De Borne 1997). All studies included an independent individual who was blinded to control and test groups to evaluate and analyse the data. We classified 17 studies as having high risk of bias because they were described as open‐label studies, and because the outcome assessor was not blinded (Barden 2013; Buckman 2015; Chen 1986; Fantin 2016; Fazio 2004; Foppa 2002; Hering 2011; Kawano 1992; Kawano 2000; Koenig 1997; Kojima 1993; Maufrais 2017; Rossinen 1997; Stott 1987; Stott 1991; Williams 2004; Zeichner 1985). One study ‐ Nishiwaki 2017 (a single‐blinded study) ‐ ensured participant blinding but not blinding of outcome assessors. We classified five studies as having uncertain risk of detection bias. Karatzi 2005, Mahmud 2002, Maule 1993, and Potter 1986 did not mention the method of blinding of outcome assessors. Even though Dumont 2010 mentioned blinding of outcome assessors, it is not clear whether blinding of outcome assessment was maintained in the case of blood pressure and heart rate measurements.

Incomplete outcome data

We classified 18 studies as having low risk of attrition bias (Agewall 2000; Barden 2013; Bau 2005; Bau 2011; Dai 2002; Dumont 2010; Fantin 2016; Foppa 2002; Karatzi 2013; Kawano 1992; Kojima 1993; Mahmud 2002; Narkiewicz 2000; Potter 1986; Rossinen 1997; Stott 1987; Stott 1991; Williams 2004). Dumont 2010, Karatzi 2013, Kawano 1992, and Williams 2004 reported reasons for participant withdrawal and excluded their data from the final analysis. Data were balanced across groups, hence missing data did not affect the final results.

We classified one study as having high risk of bias. Chen 1986 reported that two participants in the alcohol group dropped out of the study for unknown reasons, so data analyses were based on eight participants in the alcohol group and on 10 participants in the control group. Because the reasons behind withdrawal were not mentioned in this study, we considered this study to have high risk of bias.

We classified 13 studies as having uncertain risk of attrition bias because study authors did not explicitly mention whether all participants were included in the final analysis (Buckman 2015; Cheyne 2004; Fazio 2004; Hering 2011; Karatzi 2005; Kawano 2000; Koenig 1997; Maufrais 2017; Maule 1993; Nishiwaki 2017; Rosito 1999; Van De Borne 1997; Zeichner 1985).

Selective reporting

We (ST and CT) assessed reporting bias based on four categories: selective reporting of systolic blood pressure (SBP), selective reporting of diastolic blood pressure (DBP), selective reporting of mean arterial blood pressure (MAP), and selective reporting of heart rate (HR).

In the case of selective reporting of systolic blood pressure (SBP), we classified 25 studies as having low risk of bias because they recorded and reported SBP in the Results section or in the Figure (Barden 2013; Bau 2005; Buckman 2015; Chen 1986; Cheyne 2004; Dai 2002; Fantin 2016; Foppa 2002; Hering 2011; Karatzi 2005; Kawano 1992; Kawano 2000; Koenig 1997; Kojima 1993; Mahmud 2002; Maule 1993; Narkiewicz 2000; Nishiwaki 2017; Potter 1986; Rosito 1999; Rossinen 1997; Stott 1987; Stott 1991; Williams 2004; Zeichner 1985). We classified seven studies as having high risk of bias (Agewall 2000; Bau 2011; Dumont 2010; Fazio 2004; Karatzi 2013; Maufrais 2017; Van De Borne 1997). Agewall 2000 measured blood pressure upon arrival of participants and did not measure blood pressure after the intervention. The aim of Bau 2011 was to determine the effects of alcohol on heart rate variability, so SBP was not measured in this study. Dumont 2010 measured blood pressure during the study period, but study authors did not provide the before and after measurement of SBP. They mentioned only that change in blood pressure was not significant. The aim of Fazio 2004 was to determine the effects of alcohol on blood flow volume and velocity. Blood pressure was also measured but was not reported. Study authors mentioned only that acute ethanol administration caused a transitory increase in BP at 20 minutes. Karatzi 2013 Maufrais 2017 and Van De Borne 1997 measured blood pressure before and after treatment but did not report these measurements.

In the case of selective reporting of diastolic blood pressure (DBP), we classified 23 studies as having low risk of bias because they measured and reported DBP in the Results section or in the figure (Barden 2013; Bau 2005; Chen 1986; Cheyne 2004; Dai 2002; Fantin 2016; Foppa 2002; Hering 2011; Karatzi 2005; Kawano 1992; Kawano 2000; Koenig 1997; Kojima 1993; Mahmud 2002; Maule 1993; Narkiewicz 2000; Nishiwaki 2017; Potter 1986; Rosito 1999; Stott 1987; Stott 1991; Williams 2004; Zeichner 1985). We classified nine studies as having high risk of bias (Agewall 2000; Bau 2011; Buckman 2015; Dumont 2010; Fazio 2004; Karatzi 2013; Maufrais 2017; Rossinen 1997; Van De Borne 1997). Agewall 2000 measured blood pressure upon participants' arrival and did not measure blood pressure after the intervention. The aim of Bau 2011 was to determine the effects of alcohol on heart rate variability, so study authors did not measure and report DBP. For Buckman 2015, blood pressure was recorded beat to beat continuously, but DBP was not reported. Dumont 2010 measured blood pressure during the RCT, but study authors did not provide the before and after measurement of DBP. They mentioned that changes in blood pressure were not significant. The aim of Fazio 2004 was to determine effects of alcohol on blood flow volume and velocity. Blood pressure was also measured but was not reported. Study authors mentioned that acute ethanol administration caused transitory increase in BP at 20 minutes. Rossinen 1997 measured blood pressure but selectively reported only SBP instead of reporting both SBP and DBP. Karatzi 2013 Maufrais 2017 and Van De Borne 1997 measured blood pressure before and after treatment but did not report these measurements.

For selective reporting of mean arterial blood pressure (MAP), we classified 11 studies as having low risk of bias because MAP was measured and reported (Buckman 2015; Dumont 2010; Fazio 2004; Foppa 2002; Karatzi 2005; Karatzi 2013; Kojima 1993; Maufrais 2017; Maule 1993; Narkiewicz 2000; Van De Borne 1997). We classified 21 studies as having high risk of bias because they did not report MAP (Agewall 2000; Barden 2013; Bau 2005; Bau 2011; Chen 1986; Cheyne 2004; Dai 2002; Fantin 2016; Hering 2011; Kawano 1992; Kawano 2000; Koenig 1997; Mahmud 2002; Nishiwaki 2017; Potter 1986; Rosito 1999; Rossinen 1997; Stott 1987; Stott 1991; Williams 2004; Zeichner 1985).

For selective reporting for heart rate (HR), we classified only Koenig 1997 as having high risk of bias because heart rate was not reported. We classified the remaining 33 studies as having low risk of bias because heart rate was measured and reported.

Other potential sources of bias

We divided other potential sources of bias into two main categories: conflict of interest/industry sponsorship and registration with clinical trials.gov.

In the case of assessing conflicts of interest and industry sponsorship, we classified 21 studies as having low risk of bias because they reported industry sponsorship and had no conflicts of interest (Agewall 2000; Barden 2013; Bau 2005; Bau 2011; Buckman 2015; Cheyne 2004; Dai 2002; Dumont 2010; Fantin 2016; Hering 2011; Karatzi 2013; Kawano 1992; Kawano 2000; Kojima 1993; Maufrais 2017; Narkiewicz 2000; Nishiwaki 2017; Potter 1986; Van De Borne 1997; Williams 2004; Zeichner 1985). We classified 11 studies as having uncertain risk of bias because the funding source or conflicts of interest were not reported (Chen 1986; Fazio 2004; Foppa 2002; Karatzi 2005; Koenig 1997; Mahmud 2002; Maule 1993; Rosito 1999; Rossinen 1997; Stott 1987; Stott 1991).

In the case of registration at clinical trials.gov, we considered only one study to have low risk of bias (Barden 2013). The trial was registered with the Australian New Zealand Clinical Trials Registry (ANZCTR). We classified the remaining studies as having high risk of bias because the protocol was not registered and the study identifier was not reported. Therefore, it is difficult to determine a priori selection of primary and secondary outcome measures for the included studies.

Publication bias

We did not identify enough studies to construct a funnel plot for the outcomes under low doses of alcohol. We interpreted only funnel plots that were constructed based on studies reporting outcomes under medium dose and high dose of alcohol versus placebo comparisons.

We created a funnel plot using the mean difference (MD) from studies reporting effects of medium doses and high doses of alcohol on SBP, DBP, MAP, and HR against standard error (SE) of the MD to check for the existence of publication bias. Visual inspection of funnel plots shows that the effect estimate is equally distributed around the mean in Figure 4, Figure 5, Figure 6. Figure 7, and Figure 8. In Figure 9, Figure 10, and Figure 11, we observed slight asymmetry in the funnel plot that was probably due to heterogeneity rather than to publication bias. We noted some overlap of data points in some funnel plots, indicating that some of the included studies were of similar size. According to Chapter 10 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), a funnel plot asymmetry test should not be used if all studies are of similar size. In this review, most of the included studies are of similar size.

Figure 4

Funnel plot of comparison: 2 Medium‐dose alcohol vs placebo, outcome: 2.4 Heart rate.

Figure 5

Funnel plot of comparison: 3 High‐dose alcohol vs placebo, outcome: 3.1 Systolic blood pressure.

Figure 6

Funnel plot of comparison: 3 High‐dose alcohol vs placebo, outcome: 3.2 Diastolic blood pressure.

Figure 7

Funnel plot of comparison: 3 High‐dose alcohol vs placebo, outcome: 3.3 Mean arterial blood pressure.

Figure 8

Funnel plot of comparison: 3 High‐dose alcohol vs placebo, outcome: 3.4 Heart rate.

Figure 9

Funnel plot of comparison: 2 Medium‐dose alcohol vs placebo, outcome: 2.1 Systolic blood pressure.

Figure 10

Funnel plot of comparison: 2 Medium‐dose alcohol vs placebo, outcome: 2.2 Diastolic blood pressure.

Figure 11

Funnel plot of comparison: 2 Medium‐dose alcohol vs placebo, outcome: 2.3 Mean arterial blood pressure.

Effects of interventions

See: Summary of findings 1 Effect of high‐dose alcohol compared to placebo ; Summary of findings 2 Effect of medium‐dose alcohol compared to placebo; Summary of findings 3 Effect of low‐dose alcohol compared to placebo

Low dose

A dose of 14 grams of pure alcohol/ethanol or less was defined as a low dose of alcohol.

Effects of low doses of alcohol on SBP

≤ 6 hours after alcohol consumption: based on two studies in 28 participants (Cheyne 2004; Nishiwaki 2017), the mean decrease in SBP with a fixed‐effect model was 1.46 mmHg (95% confidence interval (CI) ‐8.38 to 5.42; P = 0.67). There was no heterogeneity (I² = 0%)
7 to 12 hours after alcohol consumption: unfortunately, none of the included studies reported results at this time interval
≥ 13 hours of alcohol consumption: unfortunately, none of the included studies reported results at this time interval.

Effects of low doses of alcohol on DBP

≤ 6 hours after alcohol consumption: two studies reported the early effect of alcohol consumption on DBP; the decrease in DBP was 1.46 mmHg (95% CI ‐6.91 to 3.99; P = 0.36) (Cheyne 2004; Nishiwaki 2017). There was no heterogeneity (I² = 0%)
7 to 12 hours after alcohol consumption: unfortunately, none of the included studies reported data at this time interval
≥ 13 hours of alcohol consumption: unfortunately, none of the included studies reported data at this time interval

Effects of low doses of alcohol on MAP

≤ 6 hours after alcohol consumption: based on MAP data from two studies (Cheyne 2004; Nishiwaki 2017), the mean decrease in DBP for MAP was 1.45 mmHg (95% CI ‐4.55 to 1.65; P = 0.36). There was no heterogeneity (I² = 0%)
7 to 12 hours after alcohol consumption: unfortunately, none of the included studies reported data at this time interval
≥ 13 hours of alcohol consumption: unfortunately, none of the included studies reported data at this time interval

Effects of low doses of alcohol on HR

≤ 6 hours after alcohol consumption: based on the early effect of a low dose of alcohol on HR data from two studies (Cheyne 2004; Nishiwaki 2017), HR was increased significantly by 5.06 bpm (95% CI 1.88 to 8.24; P = 0.002). There was no heterogeneity (I² = 0%)
7 to 12 hours after alcohol consumption: unfortunately, none of the included studies reported data at this time interval
≥ 13 hours after alcohol consumption: unfortunately, none of the included studies reported data at this time interval

Medium dose

Effects of medium doses of alcohol on SBP

≤ 6 hours after alcohol consumption: based on data from nine studies in 149 participants (Chen 1986; Fantin 2016; Foppa 2002; Karatzi 2005; Kawano 1992; Kawano 2000; Kojima 1993; Nishiwaki 2017; Rosito 1999), a medium dose of alcohol decreased SBP by 5.63 mmHg (95% CI ‐8.25 to ‐3.02; P < 0.001). The I² statistic value was 63% and the P value for the Chi² test was 0.004, indicating substantial heterogeneity across studies.
- Due to the presence of substantial heterogeneity across studies, we also conducted this analysis using the random‐effects model. A medium dose of alcohol decreased SBP by 6.15 mmHg (95% CI ‐10.55 to ‐1.75; P = 0.006)
7 to 12 hours after consumption: four studies in 54 participants showed that a medium dose of alcohol decreased SBP by 3.22 mmHg (95% CI ‐8.37 to 1.93; P = 0.22) 7 to 12 hours after consumption of a medium dose of alcohol (Fantin 2016; Foppa 2002; Kawano 1992; Kawano 2000). There was no heterogeneity (I² = 0)
≥ 13 hours after consumption: based on data for 66 participants from four studies (Foppa 2002; Kawano 1992; Kawano 2000; Rosito 1999), the mean increase in SBP was 0.64 mmHg (95% CI ‐3.90 to 5.18; P = 0.78). There was no heterogeneity (I² = 0)

Effects of medium doses of alcohol on DBP

≤ 6 hours after alcohol consumption: based on data from nine studies in 149 participants (Chen 1986; Fantin 2016; Foppa 2002; Karatzi 2005; Kawano 1992; Kawano 2000; Kojima 1993; Nishiwaki 2017; Rosito 1999), a medium dose of alcohol decreased DBP by 4.01 mmHg (95% CI ‐6.02 to ‐2.00; P < 0.001). The I² statistic value was 59% and the P value for the Chi² test was 0.009, indicating substantial heterogeneity across studies.
- Due to the presence of substantial heterogeneity across studies, we also conducted this analysis using the random‐effects model. A medium dose of alcohol decreased DBP by 3.76 mmHg (95% CI ‐7.02 to ‐0.50; P = 0.006)
7 to 12 hours after alcohol consumption: four studies in 54 participants reported a mean decrease in DBP of 1.19 mmHg (95% CI ‐4.29 to 1.90; P = 0.45) after 7 to 12 hours after consumption of a medium dose of alcohol (Fantin 2016; Foppa 2002; Kawano 1992; Kawano 2000). There was no heterogeneity (I² = 0)
≥ 13 hours after alcohol consumption: Based on data from 66 participants in four studies (Foppa 2002; Kawano 1992; Kawano 2000; Rosito 1999), the mean increase in DBP was 1.78 mmHg (95% CI ‐0.95 to 4.51; P = 0.64). There was no heterogeneity (I² = 0)

Effects of medium doses of alcohol on MAP

≤ 6 hours after alcohol consumption: based on data from 17 studies in 428 participants (Barden 2013; Bau 2005; Buckman 2015; Dai 2002; Dumont 2010; Hering 2011; Koenig 1997; Mahmud 2002; Maule 1993; Narkiewicz 2000; Potter 1986; Rosito 1999; Rossinen 1997; Stott 1987; Stott 1991; Williams 2004; Zeichner 1985), we found that a high dose of alcohol decreased MAP by 1.53 mmHg (95% CI ‐3.34 to 0.28; P = 0.10). There was evidence of low heterogeneity across studies (I² = 27%)
- Due to the presence of substantial heterogeneity across studies, we also conducted this analysis using the random‐effects model. A medium dose of alcohol decreased MAP by 2.92 mmHg (95% CI –5.76 to ‐0.07; P = 0.04)
7 to 12 hours after alcohol consumption: based on data from four studies in 54 participants (Fantin 2016; Foppa 2002; Kawano 1992; Kawano 2000), the mean decrease in MAP was 2.11 mmHg (95% CI ‐4.69 to 0.48; P = 0.11) after 7 to 12 hours after consumption of a medium dose of alcohol. There was no heterogeneity (I² = 0)
≥ 13 hours after alcohol consumption: based on data from 66 participants from four studies (Foppa 2002; Kawano 1992; Kawano 2000; Rosito 1999), the mean increase in MAP was 1.43 mmHg (95% CI ‐1.18 to 4.04; P = 0.28). There was no heterogeneity (I² = 0)

Effects of medium doses of alcohol on HR

≤ 6 hours after alcohol consumption: 12 studies reported the effects of alcohol consumption in 181 participants (Agewall 2000; Chen 1986; Fantin 2016; Fazio 2004; Foppa 2002; Karatzi 2005; Karatzi 2013; Kawano 1992; Kawano 2000; Kojima 1993; Maufrais 2017; Nishiwaki 2017); the mean increase in HR was 4.62 bpm (95% CI 3.14 to 6.11; P < 0.01). There was no heterogeneity (I² = 0)
7 to 12 hours after alcohol consumption: based on data from four studies in 54 participants (Fantin 2016; Foppa 2002; Kawano 1992; Kawano 2000), the mean increase in HR was 1.22 bpm (95% CI –1.88 to 4.32; P = 0.44) after 7 to 12 hours after consumption of a medium dose of alcohol. There was no heterogeneity (I² = 0)
≥ 13 hours after alcohol consumption: based on three studies in 36 participants (Foppa 2002; Kawano 1992; Kawano 2000), the mean increase in HR was 1.37 bpm (95% CI ‐2.12 to 4.86; P = 0.44)

High dose

Effects of high doses of alcohol on SBP

≤ 6 hours after alcohol consumption: based on data from 418 participants in 16 studies (Barden 2013; Bau 2005; Buckman 2015; Dai 2002; Hering 2011; Koenig 1997; Mahmud 2002; Maule 1993; Narkiewicz 2000; Potter 1986; Rosito 1999; Rossinen 1997; Stott 1987; Stott 1991; Williams 2004; Zeichner 1985), high doses of alcohol decreased SBP by 3.49 mmHg (95% CI ‐6.03 to ‐0.95; P = 0.007). The I² statistic value was 45% and the P‐value for the Chi² test was 0.02, indicating moderate heterogeneity across studies
7 to 12 hours after alcohol consumption: based on data from 54 participants in three studies (Barden 2013; Rosito 1999; Stott 1987), high doses of alcohol decreased SBP by 3.72 mmHg (95% CI ‐6.97 to ‐0.48; P = 0.02). There was no heterogeneity (I² = 0)
≥ 13 hours after alcohol consumption: based on data from 154 participants in four studies (Barden 2013; Bau 2005; Rosito 1999; Stott 1987), high doses of alcohol increased SBP by 3.69 mmHg (95% CI 2.33 to 5.05; P < 0.001]. There was no heterogeneity (I² = 0)

Effects of high doses of alcohol on DBP

≤ 6 hours after alcohol consumption: based on data from 350 participants in 14 studies (Barden 2013; Bau 2005; Dai 2002; Hering 2011; Koenig 1997; Mahmud 2002; Maule 1993; Narkiewicz 2000; Potter 1986; Rosito 1999 ; Stott 1987; Stott 1991; Williams 2004; Zeichner 1985), high doses of alcohol decreased DBP by 1.91 mmHg (95% CI ‐3.86 to 0.04; P = 0.05). The I² statistic value was 35% and the P‐value for the Chi² test was < 0.1, indicating low heterogeneity across studies
7 to 12 hours after alcohol consumption: based on data from 54 participants in three studies (Barden 2013; Rosito 1999; Stott 1987), high doses of alcohol decreased DBP by 1.71 mmHg (95% CI –4.59 to 1.77; P = 0.24). There was no heterogeneity (I² = 0)
≥ 13 hours after alcohol consumption: based on data from 154 participants in four studies (Barden 2013; Bau 2005; Rosito 1999; Stott 1987), high doses of alcohol increased DBP by 2.37 mmHg (95% CI 0.24 to 4.49; P = 0.03) after 13 hours of consumption. There was no heterogeneity (I² = 0)

Effects of high doses of alcohol on MAP

≤ 6 hours after alcohol consumption: based on data from 16 studies in 418 participants (Barden 2013; Bau 2005; Buckman 2015; Dai 2002; Dumont 2010; Hering 2011; Koenig 1997; Mahmud 2002; Maule 1993; Narkiewicz 2000; Potter 1986; Rosito 1999; Stott 1987; Stott 1991; Williams 2004; Zeichner 1985), high doses of alcohol decreased MAP by 1.79 mmHg (95% CI ‐3.72 to 0.13; P = 0.07). There was evidence of low heterogeneity across studies (I² = 30%)
7 to 12 hours after alcohol consumption: three studies reported the effects of high doses of alcohol consumption in 54 participants (Barden 2013; Rosito 1999; Stott 1987); the mean decrease in MAP was 2.47 mmHg (95% CI ‐5.69 to 0.75; P = 0.13). There was no heterogeneity (I² = 0)
≥ 13 hours after alcohol consumption: four studies reported the effects of high doses of alcohol in 154 participants (Barden 2013; Bau 2005; Rosito 1999; Stott 1987); the mean increase in MAP was 2.96 mmHg (95% CI 0.35 to 5.58, P = 0.02) ≥ 13 hours after consumption. There was no heterogeneity (I² = 0)

Effects of high doses of alcohol on HR

≤ 6 hours after alcohol consumption: 17 studies reported the effects of high doses of alcohol in 495 participants (Barden 2013; Bau 2005; Bau 2011; Buckman 2015; Dai 2002; Dumont 2010; Hering 2011; Maule 1993; Narkiewicz 2000; Potter 1986; Rosito 1999; Rossinen 1997; Stott 1987; Stott 1991; Van De Borne 1997; Williams 2004; Zeichner 1985); the mean increase in HR was 5.75 bpm (95% CI 3.99 to 7.51; P < 0.001). There was evidence of low heterogeneity across studies (I² = 34%)
7 to 12 hours after alcohol consumption: five studies reported the effects of high doses of alcohol in 144 participants (Barden 2013; Bau 2011; Rosito 1999; Rossinen 1997; Stott 1987); the mean increase in HR was 6.16 bpm (95% CI 3.04 to 9.28; P < 0.001). There was moderate heterogeneity across studies (I² = 44%)
≥ 13 hours after alcohol consumption: six studies reported the effects of high doses of alcohol in 244 participants (Barden 2013; Bau 2005; Bau 2011; Rosito 1999; Rossinen 1997; Stott 1987); the mean increase in HR was 2.70 bpm (95% CI 0.80 to 4.60; P = 0.005) ≥ 13 hours after consumption. There was no heterogeneity (I² = 0)

The above‐mentioned results are summarised in summary of findings Table 1, summary of findings Table 2, and summary of findings Table 3.

Subgroup analysis

It is recommended that there should be at least 10 studies reporting each of the subgroups in question. Among the 32 included studies, only four studies included hypertensive participants (Kawano 1992; Kawano 2000; Kojima 1993; Foppa 2002). So, it was not appropriate to conduct a separate meta‐analysis based on that population.

For the planned subgroup analysis based on sex, no studies reported male and female participant data separately. Therefore, we were unable to perform a subgroup analysis based on the sex of participants.

Sensitivity analysis

As planned, we conducted sensitivity analyses to see if there was any significant difference between effect estimates of outcomes given by the fixed‐effect model and the random‐effects model, when substantial heterogeneity was present. The result is presented in Table 3; there was no significant difference between results given by the two models.

Open in table viewer

Table 3. Sensitivity analysis: fixed‐effect model vs random‐effects model

Outcomes or subgroup	Mean difference, IV, fixed‐effect model, 95% CI	Mean difference, IV, random‐effects model, 95% CI
2.1 SBP (≤ 6 hours)	‐5.63 [‐8.25, ‐3.02] Heterogeneity: Chi² = 24.51, df = 9 (P = 0.004); I² = 63% Test for overall effect: Z = 4.22 (P < 0.0001)	‐6.15 [‐10.55, ‐1.75] Heterogeneity: Chi² = 24.51, df = 9 (P = 0.004); I² = 63% Test for overall effect: Z = 2.74 (P = 0.006)
2.2 DBP (≤ 6 hours)	‐4.01 [‐6.02, ‐2.00] Heterogeneity: Chi² = 21.81, df = 9 (P = 0.009); I² = 59% Test for overall effect: Z = 3.91 (P < 0.0001)	‐3.76 [‐7.02, ‐0.50] Heterogeneity: Chi² = 21.81, df = 9 (P = 0.009); I² = 59% Test for overall effect: Z = 2.26 (P = 0.02)
2.3 MAP (≤ 6 hours)	‐2.17 [‐3.68, ‐0.65] Heterogeneity: Chi² = 38.36, df = 12 (P = 0.0001); I² = 69% Test for overall effect: Z = 2.80 (P = 0.005)	‐2.92 [‐5.76, ‐0.07] Heterogeneity: Chi² = 38.36, df = 12 (P = 0.0001); I² = 69% Test for overall effect: Z = 2.01 (P = 0.04)

CI: confidence interval.
DBP: diastolic blood pressure.
IV: inverse variance.
MAP: mean arterial pressure.
SBP: systolic blood pressure.

We planned on conducting sensitivity analyses on studies based on their level of risk of bias (high‐risk studies versus low‐risk studies). Most of the included studies had similar risk of bias across all domains except for performance bias and detection bias, for which risk arises from blinding of participants, personnel, and outcome assessors. So, we decided to conduct a sensitivity analysis of the included studies based on the blinding condition (Table 4). We observed a greater reduction in blood pressure after a moderate dose of alcohol consumption for the unblinded studies, which was probably due to the presence of a heterogeneous population. For high‐dose alcohol studies, we did not find any significant difference between blinded and unblinded studies. Overall, the of studies did not influence outcomes.

Open in table viewer

Table 4. Sensitivity analysis: blinded studies vs unblinded studies

Outcomes or subgroups

Blinded studies, mean difference, IV, fixed‐effect model, 95% CI

Unblinded, mean difference, IV, fixed‐effect model, 95% CI

2.1 SBP (< 6 hours)

‐4.56 [‐8.39, ‐0.73]

Heterogeneity: Chi² = 2.14, df = 3 (P = 0.54); I² = 0%

Test for overall effect: Z = 2.33 (P = 0.02)

‐6.57 [‐10.15, ‐3.00]

Heterogeneity: Chi² = 21.80, df = 5 (P = 0.0006); I² = 77%

Test for overall effect: Z = 3.60 (P = 0.0003)

2.2 DBP (< 6 hours)

‐1.50 [‐4.89, 1.89]

Heterogeneity: Chi² = 1.99, df = 3 (P = 0.57); I² = 0%

Test for overall effect: Z = 0.86 (P = 0.39)

‐5.37 [‐7.86, ‐2.87]

Heterogeneity: Chi² = 16.57, df = 5 (P = 0.005); I² = 70%

Test for overall effect: Z = 4.22 (P < 0.0001)

2.3 MAP (< 6 hours)

‐0.11 [‐3.39, 3.18]

Heterogeneity: Chi² = 8.24, df = 4 (P = 0.08); I² = 51%

Test for overall effect: Z = 0.06 (P = 0.95)

‐4.93 [‐8.83, ‐1.02]

Heterogeneity: Chi² = 21.61, df = 7 (P = 0.003); I² = 68%

Test for overall effect: Z = 2.47 (P = 0.01)

2.4 HR (< 6 hours)

4.35 [2.31, 6.40]

Heterogeneity: Chi² = 2.33, df = 3 (P = 0.51); I² = 0%

Test for overall effect: Z = 4.17 (P < 0.0001)

4.92 [2.77, 7.08]

Heterogeneity: Chi² = 7.37, df = 7 (P = 0.39); I² = 5%

Test for overall effect: Z = 4.48 (P < 0.00001)

3.1 SBP (< 6 hours)

‐3.80 [‐8.03, 0.43]

Heterogeneity: Chi² = 21.01, df = 7 (P = 0.004); I² = 67%

Test for overall effect: Z = 1.76 (P = 0.08)

‐2.84 [‐5.53, ‐0.14]

Heterogeneity: Chi² = 6.04, df = 7 (P = 0.54); I² = 0%

Test for overall effect: Z = 2.06 (P = 0.04)

3.2 DBP (< 6 hours)

‐1.88 [‐4.73, 0.97]

Heterogeneity: Tau² = 6.68; Chi² = 11.57, df = 6 (P = 0.07); I² = 48%

Test for overall effect: Z = 1.29 (P = 0.20)

‐1.99 [‐4.89, 0.90]

Heterogeneity: Tau² = 3.87; Chi² = 8.16, df = 6 (P = 0.23); I² = 26%

Test for overall effect: Z = 1.35 (P = 0.18)

3.3 MAP (< 6 hours)

‐1.62 [‐3.98, 0.74]

Heterogeneity: Chi² = 10.45, df = 8 (P = 0.23); I² = 23%

Test for overall effect: Z = 1.35 (P = 0.18)

‐1.44 [‐4.46, 1.57]

Heterogeneity: Chi² = 11.54, df = 7 (P = 0.12); I² = 39%

Test for overall effect: Z = 0.94 (P = 0.35)

3.4 HR (< 6 hours)

4.82 [3.01, 6.63]

Heterogeneity: Chi² = 5.79, df = 8 (P = 0.67); I² = 0%

Test for overall effect: Z = 5.22 (P < 0.00001)

6.62 [3.21, 10.03]

Heterogeneity: Chi² = 16.68, df = 7 (P = 0.02); I² = 58%

Test for overall effect: Z = 3.81 (P = 0.0001)

CI: confidence interval.
DBP: diastolic blood pressure.
HR: heart rate.
IV: inverse variance.
MAP: mean arterial pressure.
SBP: systolic blood pressure.

Discussion

This review summarises the acute effects of different doses of alcohol on blood pressure and heart rate in adults (≥ 18 years of age) during three different time intervals after ingestion of alcohol.

Summary of main results

Effects of low‐dose alcohol consumption

Low‐dose alcohol consumption had no effect on blood pressure (BP) within six hours, but we found only two trials that studied this dose and no trials that assessed BP after six hours. Low‐dose alcohol increased heart rate (HR) within six hours, suggesting that even one glass of wine increases HR. Unfortunately, we found no studies measuring HR more than six hours after the dose.

Effects of medium‐dose alcohol consumption

Medium‐dose alcohol decreased systolic blood pressure (SBP) by 5.6 mmHg and diastolic blood pressure (DBP) by 4 mmHg within the first six hours of consumption. Although the hypotensive effect of alcohol seemed to last up to 12 hours after drinking alcohol, and the effect was lost after 13 hours, the result was based on only four trials reporting intermediate (7 to 12 hours) and late (after 13 hours) effects of alcohol on BP.

Heart rate was increased by 4.6 bpm six hours after drinking alcohol compared to placebo. Intermediate (7 to 12 hours) and late (after 13 hours) effects of the medium dose of alcohol on HR were based on only four trials and were not statistically different compared to placebo.

Effects of high‐dose alcohol consumption

High‐dose alcohol decreased SBP by 3.49 mmHg within the first six hours, and by 3.77 mmHg between 7 and 12 hours after consumption. After 13 hours, high doses of alcohol increased SBP by 3.7 mmHg compared to placebo. DBP was not significantly affected up to 12 hours after drinking a high dose of alcohol, but there was a statistically significant increase in DBP during the ≥ 13 hour time interval after alcohol consumption.

High‐dose alcohol consumption increased HR by approximately 6 bpm in participants, and the effect lasted up to 12 hours. After that, HR was still raised in participants, but it averaged 2.7 bpm.

Dose‐dependent response

There is likely a dose‐response effect of alcohol on BP, as the effects of alcohol appeared to last longer with higher doses. However, we lacked data on the effects of low doses. We intended to find out the dose‐dependent changes in SBP, DBP, mean arterial pressure (MAP), and HR after consumption of a single dose of alcohol. Because the numbers of included studies that fell into our pre‐specified dose categories were not comparable, we were unable to conduct a comprehensive dose‐dependent analysis. Rosito 1999 tested the effects of 15 g, 30 g, and 60 g of alcohol on 40 young medical students. The decrease in SBP was greater with 30 g of alcohol seven hours after consumption compared to placebo and 15 g and 60 g alcohol‐consuming groups. In this study, alcohol had no significant effect on DBP in the four groups.

Possible mechanisms to explain the results

Many interrelated changes are possibly responsible for the biphasic effect of alcohol on blood pressure.

Acute alcohol consumption was found to reduce 20‐hydroxyeicosatetraenoic acid (20‐HETE) in Barden 2013. 20‐HETE is a signalling molecule with a wide range of effects on the cardiovascular system. It is a vasoconstrictor that inhibits sodium reabsorption in the proximal and distal tubules of the kidney. The reduction in 20‐HETE was greatest when the blood alcohol level was highest (Barden 2013; Collins 2005). Increased production of nitric oxide (NO) was also associated with acute alcohol consumption (Deng 2007; Rocha 2012). NO is released from the endothelium and is a potent vasodilator. Also, there is an inverse relationship between NO and 20‐HETE. Alcohol is often consumed mixed together with fruit juices, and the combination was found to promote insulin secretion (Steiner 2015). Insulin is known to signal a phosphorylation‐dependent mechanism resulting in the production of NO (Muniyappa 2007). Moreover, alcohol's metabolic byproduct acetaldehyde is another known vasodilator (Altura 1978; Gillespie 1967). Together, the reduction in vasoconstrictors and the increase in vasodilators could be the explanation for the drop in blood pressure, which lasted approximately 12 hours.

The blood alcohol level decreased over time, and 20‐HETE started to rise (Barden 2013). The hypertensive effect of alcohol after 13 hours of consumption could be the result of the rise in vasoconstrictors and the homeostatic response to restore blood pressure. Plasma renin activity was reported to be increased in Kawano 2000 as a late effect of alcohol consumption.

Heart rate was increased following alcohol consumption regardless of the dose of alcohol. Alcohol has been shown to slow down parasympathetic nervous activity and to stimulate sympathetic nervous activity. Hering 2011, Carter 2011, and Spaak 2008 reported an increase in muscle sympathetic nervous activity (MSNA), which persists for at least 10 hours after consumption. The vagus nerve is a component of the parasympathetic nervous system and is largely responsible for regulation of the heart rate at rest. Rossinen 1997 and Van De Borne 1997 reported withdrawal of vagal tone and reduced heart rate variability within an hour after alcohol consumption; this explains the increased heart rate. Buckman 2015, Van De Borne 1997, and Fazio 2001 also reported reduced baroreflex sensitivity following alcohol consumption. Impairment of baroreflex sensitivity results in failure to sense the increase in heart rate and maintenance of cardiovascular homeostasis. Kawano 2000 reported a reduction in plasma potassium levels after alcohol consumption, which might provide another reason for the increase in heart rate.

Clinical implications of the findings

This systematic review provides us with a better understanding of the time‐course of alcohol's acute effects on blood pressure and heart rate. This review included only short‐term randomised controlled trials (RCTs) investigating the effects of alcohol on blood pressure and heart rate. Acute alcohol consumption mimics the pattern of social drinking, and evidence indicates that even one glass of an alcoholic drink can increase heart rate. The magnitude of the effects of alcohol on blood pressure and heart rate varies, based possibly on genetic factors and on the amount of alcohol consumed.

Several long‐term observational studies have reported that light to moderate alcohol consumption is associated with a reduction in adverse cardiovascular events (Briasoulis 2012; Di Castelnuovo 2006; Plunk 2014; Taylor 2009). At the present time, the explanation for these possible beneficial effects of regular light to moderate alcohol consumption is unknown. In this review, consuming medium (moderate) doses of alcohol lowered blood pressure for possibly up to 12 hours after consumption. Perhaps regular light to moderate consumption of alcohol lowers blood pressure, and this is the explanation for the reduction in cardiovascular events seen in the observational studies.

On the other hand, this review shows that high doses of alcohol increased blood pressure after 13 hours and could lead to increased blood pressure the day after consumption. Thus regular consumption of high doses of alcohol could lead to a sustained increase in blood pressure and the adverse consequences associated with hypertension.

Hypertensive individuals who take antihypertensive drugs should be cautious about the timing of drinking alcohol because the combination of some antihypertensive drugs and alcohol was found to synergistically reduce blood pressure (Bailey 1989; Kawano 1999; Kawano 2000).

Overall completeness and applicability of evidence

The evidence synthesised in this review was collected from 32 RCTs in 767 participants. Of the 32 studies, two studied low‐dose alcohol, 12 studied medium‐dose alcohol, and 19 studied high‐dose alcohol. Rosito 1999 studied both medium and high doses of alcohol. The sample size in the meta‐analysis for low‐dose comparison was not adequate to assess the effects of low doses of alcohol on BP and HR; however, we believe that the direction of the change in BP and HR was correct. For medium doses and high doses of alcohol, participants represented a range in terms of age, sex, and health condition. Because the participant population comprised predominantly young and healthy normotensive men, the overall evidence generated in this review cannot be extrapolated to women and older populations with other comorbidities.

There was substantial heterogeneity in the meta‐analysis for effects of medium doses on SBP and DBP six hours after alcohol consumption (Analysis 2.1 and Analysis 2.2) due to the inclusion of Kawano 1992, Kawano 2000, and Kojima 1993. Participants in these studies were Japanese with mean age > 50 years and essential hypertension. The mean reduction in SBP and DBP was greater in these participants compared to participants from other studies ingesting medium doses of alcohol. The reason behind the greater reduction in blood pressure after consuming alcohol is probably the presence of genetic variants of the hormones alcohol dehydrogenase (ADH) and aldehyde dehydrogenase (ALDH) among Japanese participants. The presence of these gene variants can cause an increase in the blood acetaldehyde level to more than 10 times the normal level (Enomoto 1991; Peng 2014; Thomasson 1995). The ALDH2*2 allele is a gene variant of the enzyme ALDH and is almost exclusively present in the northeast Asian population (Wall 2016). Studies on liver extracts have reported that this particular allele of ALDH is dominant. The presence of this allele leads to reduced acetaldehyde metabolising activity in heterozygotes and no metabolising activity in homozygotes (Li 2009; Li 2012). In both cases, the blood acetaldehyde level after alcohol consumption was found to be greater than the normal range. An excessive amount of acetaldehyde in blood is responsible for facial flushing, which is a common adverse effect of alcohol consumption among east Asians. Acetaldehyde could be responsible for the hypotensive effects of alcohol (Altura 1978; Gillespie 1967), and excessive acetaldehyde could be the reason for the greater reduction in blood pressure reported in the aforementioned studies including Japanese populations.

To check whether the age of participants had any role in the effects of alcohol on blood pressure, we selected studies with older participants (> 50 years) from the analysis (Analysis 2.1; Analysis 2.2; Analysis 3.1; Analysis 3.2). Studies with older participants showed a greater decrease in SBP and DBP compared to studies with younger participants (< 50 years) (results presented in Table 5). Although we do not have much confidence in the mean change in blood pressure in the older population due to fewer studies compared to studies with younger participants and the presence of heterogeneity, previous research suggests that age can affect the rate of elimination of alcohol from the body (Cederbaum 2012; Hahn 1983). With increasing age, the amount of body water and liver mass are decreased, leading to higher blood alcohol levels. Also, the metabolising enzymes such as cytochrome P‐4502E1 and ALDH diminish with age. Older populations are more likely to take more drugs to treat existing comorbidities, and the drug‐alcohol interaction can modify alcohol metabolism (Meier 2008). All these factors can lead to increased acetaldehyde in the body and greater reduction in blood pressure.

Open in table viewer

Table 5. Differences between older and younger participants

Analysis	Older participants (mean age ≥ 50), mean difference, IV, fixed‐effect model, 95% CI	Younger participants (mean age < 50), mean difference, IV, fixed‐effect model, 95% CI
2.1 SBP (< 6 hours)	‐11.25 [‐15.63, ‐6.87] Test for overall effect: Z = 5.04 (P < 0.00001); df = 3 (P = 0.03); I² = 66%	‐2.52 [‐5.78, 0.74] Test for overall effect: Z = 1.52 (P = 0.13); df = 5 (P = 0.31); I² = 16%
2.2 DBP (<6 hours)	‐7.82 [‐11.08, ‐4.57] Test for overall effect: Z = 4.71 (P < 0.00001); df = 3 (P = 0.008); I² = 74%	‐1.66 [‐4.22, 0.89] Test for overall effect: Z = 1.28 (P = 0.20); df = 5 (P = 0.91); I² = 0%
3.1 SBP (<6 hours)	‐6.71 [‐11.23, ‐2.18] Test for overall effect: Z = 2.91 (P = 0.004); df = 3 (P = 0.92); I² = 0%	‐3.04 [‐4.87, ‐1.20]; Test for overall effect: Z = 3.24 (P = 0.001); df = 11 (P = 0.010); I² = 56%
3.2 DBP (<6 hours)	‐4.30 [‐7.32, ‐1.27] Test for overall effect: Z = 2.78 (P = 0.005); df = 2 (P = 0.25); I² = 27%	‐1.67 [‐3.33, ‐0.01] Test for overall effect: Z = 1.97 (P = 0.05); df = 10 (P = 0.13); I² = 34%

CI: confidence interval.
DBP: diastolic blood pressure.
IV: inverse variance.
SBP: systolic blood pressure.

Rosito 1999 reported the effects of 15, 30, and 60 g of alcohol compared to placebo on healthy male volunteers. According to our pre‐specified dose categories, both 15 g and 30 g of alcohol fell under the medium dose category. Including both of these doses or de‐selecting either one of these doses from Rosito 1999 from Analysis 2.1 and Analysis 2.2 (medium doses of alcohol) resulted in the same statistically significant conclusion.

We identified Stott 1987 and Barden 2013 from Analysis 3.1 and Analysis 3.2 as having a considerably lower standard error (SE) of the mean difference (MD) compared to the other included studies. Assuming that the low SEs of MDs reported in Stott 1987 and Barden 2013 are errors and are not reliable, we replaced these measures with the average SE of MD from the rest of the included studies. The statistically significant conclusions remained the same.

Nine out of the 19 studies used comparatively high (≥ 60 g or ≥ 1 g/kg) doses of alcohol compared to the 10 other studies under the high dose of alcohol category, and consumption of very high doses of alcohol in these nine trials over the reported short duration mimics the pattern of binge drinking. Hence, we conducted additional analyses to see if the very high dose of alcohol (≥ 60 g or ≥ 1 g/kg) had any dose‐related effects compared to lower high doses of alcohol (31 to 59 g of alcohol) (see Table 6). Results suggest that the decrease in BP with very high doses of alcohol is greater compared to lower high doses of alcohol. The change in heart rate was similar in both comparisons. However, the result was heterogeneous; therefore, we are unable to make any implications from this.

Open in table viewer

Table 6. Comparison between very high‐dose alcohol and lower high‐dose alcohol

Outcomes	Very high dose (≥ 60 g) Mean difference, IV, fixed‐effect model, 95% CI	Lower high dose (31 to 59 g) Mean difference, IV, fixed‐effect model, 95% CI
3.1. SBP	‐5.12 [‐7.32, ‐2.92] Test for overall effect: Z = 4.57 (P < 0.00001); df = 7 (P = 0.02); I² = 56%	‐1.20 [‐3.90, 1.49] Test for overall effect: Z = 0.88 (P = 0.38); df = 7 (P = 0.47); I² = 0%
3.2. DBP	‐3.21 [‐5.49, ‐0.92] Test for overall effect: Z = 2.75 (P = 0.006); df = 5 (P = 0.22); I² = 29%	‐1.65 [‐3.53, 0.23] Test for overall effect: Z = 1.72 (P = 0.09); df = 7 (P = 0.10); I² = 41%
3.3. MAP	2.17 [‐4.09, ‐0.25] Test for overall effect: Z = 2.21 (P = 0.03); df = 7 (P = 0.04); I² = 52%	‐0.47 [‐2.83, 1.90] Test for overall effect: Z = 0.39 (P = 0.70); df = 8 (P = 0.63); I² = 0%
3.4. HR	5.43 [3.76, 7.11] Test for overall effect: Z = 6.35 (P < 0.00001); df = 9 (P = 0.76); I² = 0%	6.09 [3.67, 8.51] Test for overall effect: Z = 4.93 (P < 0.00001); df = 6 (P = 0.005); I² = 67%

CI: confidence interval.
DBP: diastolic blood pressure.
HR: heart rate.
IV: inverse variance.
MAP: mean arterial pressure.
SBP: systolic blood pressure.

This is the first systematic review on this topic based on RCTs. Much of the current literature on alcohol does not mention the hypotensive effect of alcohol or the magnitude of change in BP or HR after alcohol consumption. This review will be useful for social and regular drinkers to appreciate the risks of low blood pressure within the first 12 hours after drinking.

Quality of the evidence

We graded the overall certainty of evidence using the GRADE approach via GRADEpro GDT software (GRADEpro 2014); we formulated summary of findings (SoF) tables.

We created three SoF tables to show the certainty of evidence and the summary of effects on outcomes of interest (SBP, DBP, and HR) for high (summary of findings Table 1), medium (summary of findings Table 2), and low doses (summary of findings Table 3) of alcohol.

Ratings of the certainty of evidence ranged from moderate to low in this review, which suggests that the effect estimates of alcohol might be slightly different than the true effects. For high doses of alcohol, we found moderate‐certainty evidence showing a decrease in SBP and low‐certainty evidence suggesting a decrease in DBP within the first six hours and 7 to 12 hours after consumption. Moderate‐certainty evidence shows that SBP and DBP rise between 13 and 24 hours after alcohol ingestion.

For medium doses of alcohol, moderate‐certainty evidence shows a decrease in SBP and DBP six hours after alcohol consumption, and low‐certainty evidence suggests a decrease in SBP and DBP for 7 to 12 hours after alcohol consumption. After ≥ 13 hours of consumption, SBP and DBP were raised; the certainty of evidence was low and medium, respectively.

For low doses of alcohol, we found low‐certainty evidence suggesting that SBP, DBP, and MAP fall within the first six hours after alcohol consumption.

We also found moderate‐certainty evidence showing that alcohol raises HR within the first six hours of consumption, regardless of the dose of alcohol. Moderate‐certainty evidence indicates an increase in heart rate after 7 to 12 hours and ≥ 13 hours after high‐dose alcohol consumption, low certainty of evidence was found for moderate dose of alcohol consumption.

We did not consider the lack of blinding of participants as a downgrading factor for certainty of evidence because we do not think that it affected the outcomes of this systematic review. Changes in blood pressure and heart rate after alcohol consumption were not the primary outcomes of interest in most of the included studies. We do not think participants were anticipating any significant influence on blood pressure or heart rate after drinking.

We also did not rate the certainty of evidence based on the funding sources of studies or on lack of a registered protocol because we did not think this would affect the effect estimates for these outcomes. However, we noted the lack of description of randomisation and allocation concealment methods in most of the included studies as a reason for downgrading because of the possibility of selection bias.

Potential biases in the review process

We faced several limitations during the review process. First, there was the possibility of undesired bias and imprecision due to imputations of missing statistics. Most of the included studies did not report the standard error (SE)/standard deviation (SD) of the mean difference (MD) for the outcomes of interest. As described in our protocol, when we were unable to obtain the required SE/SD from study authors or by calculation from the reported P value or 95% CI, we imputed data according to the pre‐specified imputation hierarchy. We most often used the reported endpoint SE/SD value to impute the SE/SD of MD. This is known to provide a good approximation of the SD of change in BP so is unlikely to lead to bias. Also, only 10 out of 32 studies reported changes in MAP after alcohol consumption along with SE/SD (Buckman 2015; Dumont 2010; Foppa 2002; Karatzi 2005; Karatzi 2013; Kojima 1993; Maufrais 2017; Maule 1993; Narkiewicz 2000; Van De Borne 1997). So, we had to calculate missing MAP values from reported SBP and DBP values using the formula mentioned in the protocol and we imputed the SE/SD for those.

Second, lack of representation of the female population was notable in the included studies. Only 16 out of 32 studies included a total of 129 (16% of total participants) female participants (Agewall 2000; Buckman 2015; Chen 1986; Cheyne 2004; Dumont 2010; Fantin 2016; Foppa 2002; Hering 2011; Mahmud 2002; Maufrais 2017; Maule 1993; Narkiewicz 2000; Rossinen 1997; Stott 1987; Stott 1991; Zeichner 1985), and this number was very low compared to 638 male participants. Only four studies included almost equal numbers of male and female participants (Buckman 2015; Foppa 2002; Maufrais 2017; Zeichner 1985). Moreover, none of the studies reported male and female data separately. As a result, we were not able to quantify the magnitude of the effects of alcohol on men and women separately. This is unfortunate, as we have reason to believe that the effects of alcohol on BP might be greater in women.

Methodological differences between studies might have affected measurement of the reported outcomes. Recent research suggests that automated ambulatory blood pressure monitors are more reliable than manual sphygmomanometers, particularly because automated monitors reduce white coat anxiety (Mirdamadi 2017). Of the 32 included studies, seven studies used a manual mercury sphygmomanometer or a semi‐automated sphygmomanometer for BP measurement (Bau 2005; Dai 2002; Karatzi 2005; Kojima 1993; Potter 1986; Rossinen 1997; Van De Borne 1997). Mixing of various measurement techniques (manual, semi‐automated, and fully automated) in the meta‐analysis might have led to some of the heterogeneity.

Another reason behind the heterogeneity was probably the variation in alcohol intake duration and in the timing of measurement of outcomes across the included studies. Most studies gave participants 15 to 30 minutes to finish their drinks, started measuring outcomes sometime after that, and continued taking measurements for a certain period, but there were some exceptions. Chen 1986 did not report consumption duration nor timing of measurement of BP and HR. Dai 2002 gave participants five minutes to consume high doses of alcohol and measured outcomes immediately. On the other hand, Fantin 2016 allowed participants to continue drinking during the period of outcome measurement. These differences in alcohol consumption duration and in outcome measurement times probably contributed to the wide variation in blood pressure in these studies and affected overall results of the meta‐analysis.

We took several steps to minimise the risk of selection bias to identify eligible studies for inclusion in the review. We used highly sensitive search strategies. We also checked the lists of references in the included studies and articles that cited the included studies in Google Scholar to identify relevant articles. Furthermore, we contacted authors of included studies to obtain all relevant data when information was insufficient or missing.

Agreements and disagreements with other studies or reviews

We are aware of one systematic review on effects of alcohol on blood pressure that was published in 2005 (McFadden 2005). McFadden 2005 included both randomised and non‐randomised studies with a minimum of 24 hours of blood pressure observation after alcohol consumption. This systematic review searched only the MEDLINE database for relevant studies, hence it was not exhaustive. Review authors included nine studies involving a total of 119 participants, and the duration of these studies was between four and seven days. Participants in those studies consumed alcohol regularly during the study period, whereas in our systematic review, we included only studies in which participants consumed alcohol for a short period. Based on nine studies, McFadden 2005 reported that the mean increase in SBP was 2.7 mmHg and in DBP was 1.4 mmHg. Only three of these studies measured BP at various time points and found that alcohol has a hypotensive effect lasting up to five hours after alcohol consumption and a hypertensive effect 20 hours after alcohol consumption that lasts until the next day. These findings are consistent with our results. However, the reported early reduction in BP was 11. 6 mmHg for SBP and 7.9 mmHg for DBP in McFadden 2005. These estimates are almost twice as large as our results. The inclusion of non‐randomised studies in McFadden 2005, which are known to be at higher risk of bias, is likely the reason for the discrepancy in the magnitude of BP effects.

Figure 1

Study flow diagram.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 3

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 4

Funnel plot of comparison: 2 Medium‐dose alcohol vs placebo, outcome: 2.4 Heart rate.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 5

Funnel plot of comparison: 3 High‐dose alcohol vs placebo, outcome: 3.1 Systolic blood pressure.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 6

Funnel plot of comparison: 3 High‐dose alcohol vs placebo, outcome: 3.2 Diastolic blood pressure.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 7

Funnel plot of comparison: 3 High‐dose alcohol vs placebo, outcome: 3.3 Mean arterial blood pressure.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 8

Funnel plot of comparison: 3 High‐dose alcohol vs placebo, outcome: 3.4 Heart rate.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 9

Funnel plot of comparison: 2 Medium‐dose alcohol vs placebo, outcome: 2.1 Systolic blood pressure.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 10

Funnel plot of comparison: 2 Medium‐dose alcohol vs placebo, outcome: 2.2 Diastolic blood pressure.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 11

Funnel plot of comparison: 2 Medium‐dose alcohol vs placebo, outcome: 2.3 Mean arterial blood pressure.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.1

Comparison 1: Low‐dose alcohol vs placebo, Outcome 1: Systolic blood pressure

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.2

Comparison 1: Low‐dose alcohol vs placebo, Outcome 2: Diastolic blood pressure

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.3

Comparison 1: Low‐dose alcohol vs placebo, Outcome 3: Mean arterial blood pressure

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.4

Comparison 1: Low‐dose alcohol vs placebo, Outcome 4: Heart rate

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 2.1

Comparison 2: Medium‐dose alcohol vs placebo, Outcome 1: Systolic blood pressure

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 2.2

Comparison 2: Medium‐dose alcohol vs placebo, Outcome 2: Diastolic blood pressure

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 2.3

Comparison 2: Medium‐dose alcohol vs placebo, Outcome 3: Mean arterial blood pressure

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 2.4

Comparison 2: Medium‐dose alcohol vs placebo, Outcome 4: Heart rate

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 3.1

Comparison 3: High‐dose alcohol vs placebo, Outcome 1: Systolic blood pressure

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 3.2

Comparison 3: High‐dose alcohol vs placebo, Outcome 2: Diastolic blood pressure

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 3.3

Comparison 3: High‐dose alcohol vs placebo, Outcome 3: Mean arterial blood pressure

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 3.4

Comparison 3: High‐dose alcohol vs placebo, Outcome 4: Heart rate

Ir a la figura de la revisiónAbrir en una pestaña nueva

Summary of findings 1. Effect of high‐dose alcohol compared to placebo

Outcomes	Participants (RCTs)	Certainty of the evidence (GRADE)	*Mean difference of high‐dose alcohol compared to placebo (95% CI)**
Effect of high‐dose alcohol compared to placebo
Patient or population: adult participants Setting: ambulatory Intervention: high‐dose alcohol (> 30 g) Comparison: placebo
Systolic blood pressure ‐ ≤ 6 hours	418 (16)	⊕⊕⊕⊝ Moderate^a	‐3.5 mmHg [‐6 to ‐0.5]
Systolic blood pressure ‐ 7 to 12 hours	54 (3)	⊕⊕⊕⊝ Moderate^a	‐3.7 mmHg [‐6.9 to ‐0.5]
Systolic blood pressure ‐ ≥ 13 hours	154 (4)	⊕⊕⊕⊝ Moderate^a	3.7 mmHg [2.3 to 5]
Diastolic blood pressure ‐ ≤ 6 hours	350 (14)	⊕⊕⊝⊝ Low^a,b	‐1.9 mmHg [‐3.9 to 0.04]
Diastolic blood pressure ‐ 7 to 12 hours	54 (5)	⊕⊕⊝⊝ Low^a,b	‐1.6 mmHg [‐4.1 to 0.9]
Diastolic blood pressure ‐ ≥ 13 hours	154 (4)	⊕⊕⊕⊝ Moderate^a	2.4 mmHg [0.3 to 4.5]
Heart rate ‐ ≤ 6 hours	495 (17)	⊕⊕⊕⊝ Moderate^a	5.5 bpm [4.3 to 6.7]
Heart rate ‐ 7 to 12 hours	144 (7)	⊕⊕⊕⊝ Moderate^a	6.2 bpm [3 to 9.3]
Heart rate ‐ ≥ 13 hours	244 (8)	⊕⊕⊕⊝ Moderate^a	2.7 bpm [0.8 to 4.6]
* The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aUnclear risk of selection bias and attrition bias in more than one study. ^b95% confidence interval around the best effect estimate includes both negligible effect and appreciable benefit.

Summary of findings 1. Effect of high‐dose alcohol compared to placebo

Ir a la tabla de la revisión

Summary of findings 2. Effect of medium‐dose alcohol compared to placebo

Outcomes	Participants (RCTs)	Certainty of the evidence (GRADE)	*Mean difference of medium‐dose alcohol compared to placebo (95% CI)**
Effect of medium‐dose alcohol compared to placebo
Patient or population: adult participants Setting: ambulatory Intervention: medium‐dose alcohol (15 to 30 g) Comparison: placebo
Systolic blood pressure ‐ ≤ 6 hours	149 (10)	⊕⊕⊕⊝ Moderate^a	‐5.63 mmHg [‐8.3 to ‐3]
Systolic blood pressure ‐ 7 to 12 hours	54 (4 )	⊕⊕⊝⊝ Low^a,b,c	‐3.2 mmHg [‐8.4 to 2]
Systolic blood pressure ‐ ≥ 13 hours	66 (5)	⊕⊕⊝⊝ Low^a,b	0.6 mmHg [‐3.9 to 5.2]
Diastolic blood pressure ‐ ≤ 6 hours	149 (10)	⊕⊕⊕⊝ Moderate^c	‐4 mmHg [‐6 to ‐2]
Diastolic blood pressure ‐ 7 to 12 hours	54 (4)	⊕⊕⊝⊝ Low^a,b	‐1.2 mmHg [‐4.3 to 1.9]
Diastolic blood pressure ‐ ≥ 13 hours	66 (5)	⊕⊕⊕⊝ Moderate^b	1.8 mmHg [‐0.9 to 4.5]
Heart rate ‐ ≤ 6 hours	181 (12)	⊕⊕⊕⊝ Moderate^c	4.6 bpm [3.1 to 6.1]
Heart rate ‐ 7 to 12 hours	54 (4)	⊕⊕⊝⊝ Low^a,b	1.2 bpm [‐1.9 to 4.3]
Heart rate ‐ > 13 hours	36 (3)	⊕⊕⊝⊝ Low^a,b	1.4 bpm [‐2.1 to 4.9]
* The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aUnclear risk of selection bias in more than one study. ^b95% confidence interval around the effect estimate includes both appreciable benefit and appreciable harm. ^cUnclear risk of selection bias and attrition bias in more than one study.

Summary of findings 2. Effect of medium‐dose alcohol compared to placebo

Ir a la tabla de la revisión

Summary of findings 3. Effect of low‐dose alcohol compared to placebo

Outcomes	Participants (RCTs)	Certainty of the evidence (GRADE)	*Mean difference of low‐dose alcohol compared to placebo (95% CI)**
Effect of low‐dose alcohol compared to placebo
Patient or population: adult participants Setting: ambulatory Intervention: low‐dose alcohol (≥ 14 g) Comparison: placebo
Systolic blood pressure ‐ ≤ 6 hours	28 (2)	⊕⊕⊝⊝ Low^a,b	‐1.9 mmHg [‐8.4 to 5.4]
Diastolic blood pressure ‐ ≤ 6 hours	28 (2 )	⊕⊕⊝⊝ Low^a,b	‐1.5 mmHg [‐6.9 to 4]
Heart rate ‐ ≤ 6 hours	28 (2)	⊕⊕⊕⊝ Moderate^a	5.1 bpm [1.88 higher to 8.24]
* The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aUnclear risk of selection bias. ^b95% confidence interval around the best effect estimate includes both negligible effect and appreciable benefit.

Summary of findings 3. Effect of low‐dose alcohol compared to placebo

Ir a la tabla de la revisión

Table 1. Baseline characteristics

Study ID	Randomised participants, N	Mean age (range)	Mean body weight, kg	Health condition	Reported dose of alcohol	Duration of intervention	Baseline SBP (SD)	Baseline DBP (SD)	Baseline HR (SD)
Agewall 2000	12	31 (younger than 40 years old)	Not reported	Healthy, normotensive non‐smokers	31.25 g	10 minutes	121 (6)	79 (4)	61 (7)
Barden 2013	24	56 (20 to 65)	Not reported	Healthy	41 g	30 minutes	115 (11)	72 (6)	62 (7)
Bau 2005	100	20.7 (18 to 25)	Not reported	Healthy non‐smokers	60 g	30 minutes	114.2	64.8	72.43 (10.9)
Bau 2011	70	20.7 (18 to 25)	Not reported	Healthy	60 g	30 minutes	Not reported	Not reported	75 (10)
Buckman 2015	72	21.5	Not reported	Healthy	0.90 mL/kg for men 0.78 mL/kg for women	15 minutes	116.9 (13.5)	Not reported	66.9 (9.9)
Chen 1986	20	19 to 32	Not reported	Healthy, normotensive non‐smokers	Target to achieve blood level of 0.05%	Not reported (mentioned that fairly fast rate)	118 (12.88)	62.25 (5.1)	63.13 (7.1)
Cheyne 2004	17	35 (21 to 46)	Not reported	Type 1 diabetes	0.35 mg/kg BW	Not reported	116.2 (18.7)	66.8 (8.2)	70 (12)
Dai 2002	40	19 to 25 years	81.35	Healthy	0.5 g/kg	5 minutes	114.1	72.6	63.5
Dumont 2010	14	22.1 (18 to 29)	Not reported	Regular user of MDMA, otherwise healthy	Target blood alcohol level 0.6%, equivalent of 2 to 3 alcoholic beverages.	3 hours to maintain target BAC	Not reported	Not reported	66
Fantin 2016	18	34.2 (25 to 53)	70.2 (53 to 85.6)	Healthy	30 g	10 hours	110.3 (12)	80 (8)	75.5 (11.5)
Fazio 2004	10	22 (20 to 25)	Not reported	Healthy	0.3 g/kg	5 minutes	Not reported	Not reported	68
Foppa 2002	13	55 (43 to 65)	Not reported	Hypertensive and centrally obese	23 g	15 minutes	130	83	72 (8.72)
Hering 2011	24	44	Not reported	13 hypertensive and 11 normotensive	1 g/kg	20 minutes	Hypertensive: 150 (21) Normotensive: 136 (13.2)	Hypertensive: 91 (14.4) Normotensive: 76 (10)	Hypertensive: 72 (7.2) Normotensive: 70 (10)
Karatzi 2005	15	52.4	Not reported	Coronary artery disease	30 g	Not reported	109.8 (9.2)	80.7 (10.8)	67.1 (13.1)
Karatzi 2013	16	28.5	77.5	Healthy non‐smokers	20 g	15 minutes	115.4 (6.2)	68.5 (5.4)	60 (8.1)
Kawano 1992	13	55.2 (22 to 70)	65.2	Mild to moderate essential hypertension	51 g	Not reported	159 (18.8)	91.3 (12)	61.5
Kawano 2000	10	54 (32 to 67)	70 (60 to 78)	Mild essential hypertension	55.3 g (1 mL/kg BW)	60 minutes	147	91	65
Koenig 1997	15	20 to 35	76	Healthy	10 mL/kg BW	30 minutes	127 (11)	80 (9.5)	Not reported
Kojima 1993	21	56.5 (33 to 73)	Not reported	Essential hypertension	1 mL/kg BW	30 minutes	146 (18.33)	89 (9.2)	59 (9.2)
Mahmud 2002	8	21 to 40	70	Healthy, normotensive non‐smokers	56 g (0.8 g/kg of BW)	10 minutes	93.3 (10)	67 (8)	Not reported
Maufrais 2017	24	23.3	62.9	Healthy	0.4 g/kg	5 minutes	Not reported	Not reported	69 (9.8)
Maule 1993	10	31 (22 to 51)	68.1 (50 to 81)	Healthy	34 g (0.5 g/kg BW)	10 minutes	122	70	62 (6.3)
Narkiewicz 2000	19	26	Not reported	Healthy	1 g/kg BW	30 minutes	111 (11.2)	61 (7.5)	57 (7.5)
Nishiwaki 2017	11	21.1 (20 to 22)	62.6	Healthy non‐smokers	11 g and 19.25 g	5 minutes	123 (6.63)	71 (6.63)	59 (9.94)
Potter 1986	16	22 (20 to 30)	77	Healthy, normotensive	0.75 g/kg	15 minutes	122.5 (11)	72.5 (9)	63.2 (8)
Rosito 1999	40	22.2 (19 to 30)	Not reported	Healthy	60 g	1 hour	124.2 (10.7)	76.2 (9.4)	70.5 (12.6)
Rossinen 1997	20	39 to 68	Not reported	Coronary artery disease and myocardial ischaemia Patients were taking usual medicine	1.25 g/kg	1 hour 30 minutes	132 (16)	Not reported	69.5
Stott 1987	10	18 to 31	56 to 101	Healthy	1.3 g/kg	1 hour	115.5	67	70.5
Stott 1991	8	81 (70 to 96)	68.4	Normotensive	0.5 g/kg	15 minutes	130 (18.3)	77.5 (15.5)	57 (7.5)
Van De Borne 1997	16	26	Not reported	Healthy	1 g/kg	30 minutes	Not reported	Not reported	59 (8)
Williams 2004	13	59 (48 to 70)	86	Coronary artery disease	0.52 g/kg	20 minutes	135 (11)	82 (7)	55 (11)
Zeichner 1985	48	20.9 (19 to 23)	Not reported	Healthy	1 g/kg	20 minutes	115.5 (10.2)	70.4 (9.1)	69.4 (12.10)
BAC: blood alcohol concentration. BW: body weight. DBP: diastolic blood pressure. HR: heart rate. SBP: systolic blood pressure. SD: standard deviation.

Table 1. Baseline characteristics

Ir a la tabla de la revisión

Table 2. Contact with corresponding authors

Study ID	Reason for contact	Contacted? (Yes/No)	Response? (Yes/No)
Agewall 2000	Method of allocation concealment used in RCT was not mentioned	No Comment ‐ contact information cannot be found	NA
Bau 2005	Method of allocation concealment used in RCT was not mentioned	Yes	No
Bau 2011	Method of allocation concealment used in RCT was not mentioned	Yes	No
Buckman 2015	Method of allocation concealment used in RCT was not mentioned	Yes	Yes
Chen 1986	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Dai 2002	Method of allocation concealment used in RCT was not mentioned	Yes	No
Dumont 2010	Methods of randomisation and allocation concealment, blinding of participants and personnel, and blinding of outcome assessment used in RCT were not mentioned	Yes	No
Fantin 2016	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Fazio 2004	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Foppa 2002	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	Yes
Hering 2011	Methods of allocation concealment used in RCT was not mentioned	Yes	Yes
Karatzi 2005	Methods of randomisation and allocation concealment, blinding of participants and personnel, and blinding of outcome assessment used in RCT were not mentioned	Yes	No
Karatzi 2013	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Kawano 1992	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes Comment ‐ contact information cannot be found in the study. However, we used contact information provided in Kawano 2000	No
Kawano 2000	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Kojima 1993	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Mahmud 2002	Methods of randomisation and allocation concealment, blinding of participants and personnel, and blinding of outcome assessment used in RCT were not mentioned	Yes	No
Maufrais 2017	Method of allocation concealment used in RCT was not mentioned	Yes	No
Maule 1993	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Narkiewicz 2000	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Nishiwaki 2017	Methods of randomisation and allocation concealment used in RCT were not mentioned	Yes	No
Potter 1986	Methods of randomisation and allocation concealment and blinding of outcome assessor used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Rossinen 1997	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Rosito 1999	Methods of randomisation and allocation concealment and blinding of participants and personnel used in RCT were not mentioned	Yes	Yes
Stott 1987	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Stott 1991	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Van De Borne 1997	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Williams 2004	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
Zeichner 1985	Methods of randomisation and allocation concealment used in RCT were not mentioned	No Comment ‐ contact information cannot be found	NA
NA: not applicable. RCT: randomised controlled trial.

Table 2. Contact with corresponding authors

Ir a la tabla de la revisión

Table 3. Sensitivity analysis: fixed‐effect model vs random‐effects model

Outcomes or subgroup	Mean difference, IV, fixed‐effect model, 95% CI	Mean difference, IV, random‐effects model, 95% CI
2.1 SBP (≤ 6 hours)	‐5.63 [‐8.25, ‐3.02] Heterogeneity: Chi² = 24.51, df = 9 (P = 0.004); I² = 63% Test for overall effect: Z = 4.22 (P < 0.0001)	‐6.15 [‐10.55, ‐1.75] Heterogeneity: Chi² = 24.51, df = 9 (P = 0.004); I² = 63% Test for overall effect: Z = 2.74 (P = 0.006)
2.2 DBP (≤ 6 hours)	‐4.01 [‐6.02, ‐2.00] Heterogeneity: Chi² = 21.81, df = 9 (P = 0.009); I² = 59% Test for overall effect: Z = 3.91 (P < 0.0001)	‐3.76 [‐7.02, ‐0.50] Heterogeneity: Chi² = 21.81, df = 9 (P = 0.009); I² = 59% Test for overall effect: Z = 2.26 (P = 0.02)
2.3 MAP (≤ 6 hours)	‐2.17 [‐3.68, ‐0.65] Heterogeneity: Chi² = 38.36, df = 12 (P = 0.0001); I² = 69% Test for overall effect: Z = 2.80 (P = 0.005)	‐2.92 [‐5.76, ‐0.07] Heterogeneity: Chi² = 38.36, df = 12 (P = 0.0001); I² = 69% Test for overall effect: Z = 2.01 (P = 0.04)
CI: confidence interval. DBP: diastolic blood pressure. IV: inverse variance. MAP: mean arterial pressure. SBP: systolic blood pressure.

Table 3. Sensitivity analysis: fixed‐effect model vs random‐effects model

Ir a la tabla de la revisión

Table 4. Sensitivity analysis: blinded studies vs unblinded studies

Outcomes or subgroups

Blinded studies, mean difference, IV, fixed‐effect model, 95% CI

Unblinded, mean difference, IV, fixed‐effect model, 95% CI

2.1 SBP (< 6 hours)

‐4.56 [‐8.39, ‐0.73]

Heterogeneity: Chi² = 2.14, df = 3 (P = 0.54); I² = 0%

Test for overall effect: Z = 2.33 (P = 0.02)

‐6.57 [‐10.15, ‐3.00]

Heterogeneity: Chi² = 21.80, df = 5 (P = 0.0006); I² = 77%

Test for overall effect: Z = 3.60 (P = 0.0003)

2.2 DBP (< 6 hours)

‐1.50 [‐4.89, 1.89]

Heterogeneity: Chi² = 1.99, df = 3 (P = 0.57); I² = 0%

Test for overall effect: Z = 0.86 (P = 0.39)

‐5.37 [‐7.86, ‐2.87]

Heterogeneity: Chi² = 16.57, df = 5 (P = 0.005); I² = 70%

Test for overall effect: Z = 4.22 (P < 0.0001)

2.3 MAP (< 6 hours)

‐0.11 [‐3.39, 3.18]

Heterogeneity: Chi² = 8.24, df = 4 (P = 0.08); I² = 51%

Test for overall effect: Z = 0.06 (P = 0.95)

‐4.93 [‐8.83, ‐1.02]

Heterogeneity: Chi² = 21.61, df = 7 (P = 0.003); I² = 68%

Test for overall effect: Z = 2.47 (P = 0.01)

2.4 HR (< 6 hours)

4.35 [2.31, 6.40]

Heterogeneity: Chi² = 2.33, df = 3 (P = 0.51); I² = 0%

Test for overall effect: Z = 4.17 (P < 0.0001)

4.92 [2.77, 7.08]

Heterogeneity: Chi² = 7.37, df = 7 (P = 0.39); I² = 5%

Test for overall effect: Z = 4.48 (P < 0.00001)

3.1 SBP (< 6 hours)

‐3.80 [‐8.03, 0.43]

Heterogeneity: Chi² = 21.01, df = 7 (P = 0.004); I² = 67%

Test for overall effect: Z = 1.76 (P = 0.08)

‐2.84 [‐5.53, ‐0.14]

Heterogeneity: Chi² = 6.04, df = 7 (P = 0.54); I² = 0%

Test for overall effect: Z = 2.06 (P = 0.04)

3.2 DBP (< 6 hours)

‐1.88 [‐4.73, 0.97]

Heterogeneity: Tau² = 6.68; Chi² = 11.57, df = 6 (P = 0.07); I² = 48%

Test for overall effect: Z = 1.29 (P = 0.20)

‐1.99 [‐4.89, 0.90]

Heterogeneity: Tau² = 3.87; Chi² = 8.16, df = 6 (P = 0.23); I² = 26%

Test for overall effect: Z = 1.35 (P = 0.18)

3.3 MAP (< 6 hours)

‐1.62 [‐3.98, 0.74]

Heterogeneity: Chi² = 10.45, df = 8 (P = 0.23); I² = 23%

Test for overall effect: Z = 1.35 (P = 0.18)

‐1.44 [‐4.46, 1.57]

Heterogeneity: Chi² = 11.54, df = 7 (P = 0.12); I² = 39%

Test for overall effect: Z = 0.94 (P = 0.35)

3.4 HR (< 6 hours)

4.82 [3.01, 6.63]

Heterogeneity: Chi² = 5.79, df = 8 (P = 0.67); I² = 0%

Test for overall effect: Z = 5.22 (P < 0.00001)

6.62 [3.21, 10.03]

Heterogeneity: Chi² = 16.68, df = 7 (P = 0.02); I² = 58%

Test for overall effect: Z = 3.81 (P = 0.0001)

CI: confidence interval.
DBP: diastolic blood pressure.
HR: heart rate.
IV: inverse variance.
MAP: mean arterial pressure.
SBP: systolic blood pressure.

Table 4. Sensitivity analysis: blinded studies vs unblinded studies

Ir a la tabla de la revisión

Table 5. Differences between older and younger participants

Analysis	Older participants (mean age ≥ 50), mean difference, IV, fixed‐effect model, 95% CI	Younger participants (mean age < 50), mean difference, IV, fixed‐effect model, 95% CI
2.1 SBP (< 6 hours)	‐11.25 [‐15.63, ‐6.87] Test for overall effect: Z = 5.04 (P < 0.00001); df = 3 (P = 0.03); I² = 66%	‐2.52 [‐5.78, 0.74] Test for overall effect: Z = 1.52 (P = 0.13); df = 5 (P = 0.31); I² = 16%
2.2 DBP (<6 hours)	‐7.82 [‐11.08, ‐4.57] Test for overall effect: Z = 4.71 (P < 0.00001); df = 3 (P = 0.008); I² = 74%	‐1.66 [‐4.22, 0.89] Test for overall effect: Z = 1.28 (P = 0.20); df = 5 (P = 0.91); I² = 0%
3.1 SBP (<6 hours)	‐6.71 [‐11.23, ‐2.18] Test for overall effect: Z = 2.91 (P = 0.004); df = 3 (P = 0.92); I² = 0%	‐3.04 [‐4.87, ‐1.20]; Test for overall effect: Z = 3.24 (P = 0.001); df = 11 (P = 0.010); I² = 56%
3.2 DBP (<6 hours)	‐4.30 [‐7.32, ‐1.27] Test for overall effect: Z = 2.78 (P = 0.005); df = 2 (P = 0.25); I² = 27%	‐1.67 [‐3.33, ‐0.01] Test for overall effect: Z = 1.97 (P = 0.05); df = 10 (P = 0.13); I² = 34%
CI: confidence interval. DBP: diastolic blood pressure. IV: inverse variance. SBP: systolic blood pressure.

Table 5. Differences between older and younger participants

Ir a la tabla de la revisión

Table 6. Comparison between very high‐dose alcohol and lower high‐dose alcohol

Outcomes	Very high dose (≥ 60 g) Mean difference, IV, fixed‐effect model, 95% CI	Lower high dose (31 to 59 g) Mean difference, IV, fixed‐effect model, 95% CI
3.1. SBP	‐5.12 [‐7.32, ‐2.92] Test for overall effect: Z = 4.57 (P < 0.00001); df = 7 (P = 0.02); I² = 56%	‐1.20 [‐3.90, 1.49] Test for overall effect: Z = 0.88 (P = 0.38); df = 7 (P = 0.47); I² = 0%
3.2. DBP	‐3.21 [‐5.49, ‐0.92] Test for overall effect: Z = 2.75 (P = 0.006); df = 5 (P = 0.22); I² = 29%	‐1.65 [‐3.53, 0.23] Test for overall effect: Z = 1.72 (P = 0.09); df = 7 (P = 0.10); I² = 41%
3.3. MAP	2.17 [‐4.09, ‐0.25] Test for overall effect: Z = 2.21 (P = 0.03); df = 7 (P = 0.04); I² = 52%	‐0.47 [‐2.83, 1.90] Test for overall effect: Z = 0.39 (P = 0.70); df = 8 (P = 0.63); I² = 0%
3.4. HR	5.43 [3.76, 7.11] Test for overall effect: Z = 6.35 (P < 0.00001); df = 9 (P = 0.76); I² = 0%	6.09 [3.67, 8.51] Test for overall effect: Z = 4.93 (P < 0.00001); df = 6 (P = 0.005); I² = 67%
CI: confidence interval. DBP: diastolic blood pressure. HR: heart rate. IV: inverse variance. MAP: mean arterial pressure. SBP: systolic blood pressure.

Table 6. Comparison between very high‐dose alcohol and lower high‐dose alcohol

Ir a la tabla de la revisión

Comparison 1. Low‐dose alcohol vs placebo

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Systolic blood pressure Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

1.1.1 ≤ 6 hours	2	56	Mean Difference (IV, Fixed, 95% CI)	‐1.48 [‐8.38, 5.42]
1.2 Diastolic blood pressure Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

1.2.1 ≤ 6 hours	2	56	Mean Difference (IV, Fixed, 95% CI)	‐1.46 [‐6.91, 3.99]
1.3 Mean arterial blood pressure Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

1.3.1 ≤ 6 hours	2	56	Mean Difference (IV, Fixed, 95% CI)	‐1.45 [‐4.55, 1.65]
1.4 Heart rate Show forest plot	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

1.4.1 ≤ 6 hours	2	56	Mean Difference (IV, Fixed, 95% CI)	5.06 [1.88, 8.24]

Comparison 1. Low‐dose alcohol vs placebo

Ir a la tabla de la revisión

Comparison 2. Medium‐dose alcohol vs placebo

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Systolic blood pressure Show forest plot	9		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

2.1.1 ≤ 6 hours	9	260	Mean Difference (IV, Fixed, 95% CI)	‐5.63 [‐8.25, ‐3.02]
2.1.2 7 to 12 hours	4	108	Mean Difference (IV, Fixed, 95% CI)	‐3.22 [‐8.37, 1.93]
2.1.3 ≥ 13 hours	4	112	Mean Difference (IV, Fixed, 95% CI)	0.64 [‐3.90, 5.18]
2.2 Diastolic blood pressure Show forest plot	9		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

2.2.1 ≤ 6 hours	9	260	Mean Difference (IV, Fixed, 95% CI)	‐4.01 [‐6.02, ‐2.00]
2.2.2 7 to 12 hours	4	108	Mean Difference (IV, Fixed, 95% CI)	‐1.19 [‐4.29, 1.90]
2.2.3 ≥ 13 hours	4	112	Mean Difference (IV, Fixed, 95% CI)	1.78 [‐0.95, 4.51]
2.3 Mean arterial blood pressure Show forest plot	12		Mean Difference (IV, Random, 95% CI)	Subtotals only

2.3.1 ≤ 6 hours	12	360	Mean Difference (IV, Random, 95% CI)	‐2.92 [‐5.76, ‐0.07]
2.3.2 7 to 12 hours	4	108	Mean Difference (IV, Random, 95% CI)	‐2.11 [‐4.69, 0.48]
2.3.3 ≥ 13 hours	4	112	Mean Difference (IV, Random, 95% CI)	1.43 [‐1.18, 4.04]
2.4 Heart rate Show forest plot	12		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

2.4.1 ≤ 6 hours	12	344	Mean Difference (IV, Fixed, 95% CI)	4.62 [3.14, 6.11]
2.4.2 7 to 12 hours	4	108	Mean Difference (IV, Fixed, 95% CI)	1.22 [‐1.88, 4.32]
2.4.3 ≥ 13 hours	3	72	Mean Difference (IV, Fixed, 95% CI)	1.37 [‐2.12, 4.86]

Comparison 2. Medium‐dose alcohol vs placebo

Ir a la tabla de la revisión

Comparison 3. High‐dose alcohol vs placebo

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Systolic blood pressure Show forest plot	16		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.1.1 ≤ 6 hours	16	620	Mean Difference (IV, Random, 95% CI)	‐3.49 [‐6.03, ‐0.95]
3.1.2 7 to 12 hours	3	88	Mean Difference (IV, Random, 95% CI)	‐3.72 [‐6.97, ‐0.48]
3.1.3 ≥ 13 hours	4	188	Mean Difference (IV, Random, 95% CI)	3.69 [2.33, 5.05]
3.2 Diastolic blood pressure Show forest plot	14		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.2.1 ≤ 6 hours	14	532	Mean Difference (IV, Random, 95% CI)	‐1.91 [‐3.86, 0.04]
3.2.2 7 to 12 hours	3	88	Mean Difference (IV, Random, 95% CI)	‐1.71 [‐4.59, 1.17]
3.2.3 ≥ 13 hours	4	188	Mean Difference (IV, Random, 95% CI)	2.37 [0.25, 4.49]
3.3 Mean arterial blood pressure Show forest plot	17		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.3.1 ≤ 6 hours	17	640	Mean Difference (IV, Random, 95% CI)	‐1.53 [‐3.34, 0.28]
3.3.2 7 to 12 hours	3	88	Mean Difference (IV, Random, 95% CI)	‐2.47 [‐5.69, 0.75]
3.3.3 ≥ 13 hours	4	188	Mean Difference (IV, Random, 95% CI)	2.96 [0.35, 5.58]
3.4 Heart rate Show forest plot	17		Mean Difference (IV, Random, 95% CI)	Subtotals only

3.4.1 ≤ 6 hours	17	704	Mean Difference (IV, Random, 95% CI)	5.75 [3.99, 7.51]
3.4.2 7 to 12 hours	5	198	Mean Difference (IV, Random, 95% CI)	6.16 [3.04, 9.28]
3.4.3 ≥ 13 hours	6	298	Mean Difference (IV, Random, 95% CI)	2.70 [0.80, 4.60]

Comparison 3. High‐dose alcohol vs placebo

Ir a la tabla de la revisión

	Idioma de la Revisión Cochrane Escoja su idioma de preferencia para las revisiones Cochrane y otros contenidos. Las secciones sin una traducción aparecerán en inglés.

	Idioma de la web Escoja su idioma de preferencia para la web de la Biblioteca Cochrane.

Idioma de la Revisión Cochrane

Idioma de la web

Resumen

Antecedentes

Objetivos

Métodos de búsqueda

Criterios de selección

Obtención y análisis de los datos

Resultados principales

Conclusiones de los autores

PICO

PICO

Population

Intervention

Comparison

Outcome

Resumen en términos sencillos

El alcohol tiene un efecto bifásico en la tensión arterial y aumenta la frecuencia cardíaca

Resumen visual

Authors' conclusions

Implications for practice

Implications for research

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Primary objectives

Secondary objective

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Additional outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

General missing data

Missing statistics

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Summary of findings and assessment of the certainty of evidence

Summary of findings and assessment of the certainty of the evidence

Results

Description of studies

Results of the search

Included studies

Excluded studies

Risk of bias in included studies

Allocation

Random sequence generation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Publication bias

Effects of interventions

Low dose

Effects of low doses of alcohol on SBP

Effects of low doses of alcohol on DBP

Effects of low doses of alcohol on MAP

Effects of low doses of alcohol on HR

Medium dose

Effects of medium doses of alcohol on SBP