Introduction

Orthostatic hypotension (OH) is defined as a sustained drop in blood pressure (BP, ≥ 20 mmHg systolic BP or ≥ 10 mmHg diastolic BP) within 3 min of standing upright [1]. It increases the risk of falls and all-cause mortality and is associated with disabling symptoms [2, 3]. It is very common, especially in older people and in those with chronic disease [4].

Neurogenic OH (nOH), a pathophysiological subtype of OH, results from central or peripheral autonomic dysfunction, leading to impairment of baroreflex-mediated vasoconstriction of skeletal muscle and splanchnic vasculature [5]. Both pharmacological and non-pharmacological treatments for nOH are poorly evidenced, and issues with efficacy, adherence and tolerability are common, creating a need for a better-quality evidence base [3].

Caffeine is a widely available, inexpensive food constituent with few side effects or associations with poor health outcomes [6, 7]. It has vasoconstrictive properties through antagonism of adenosine receptors (A1, A2A and A2B) [8] and has been shown to modestly increase BP both acutely and in the longer term in healthy individuals [9]. Low-quality evidence indicates that caffeine is an effective treatment for postprandial hypotension, another common problem in patients with autonomic failure, potentially through inhibition of adenosine-driven splanchnic vasodilatation [10]. This has led to the hypothesis that caffeine may be helpful for people with nOH. Indeed, caffeine has consequently been recommended as a treatment for refractory nOH in the literature and in clinical practice, although there is no consensus regarding its efficacy [10, 11].

Due to the significant uncertainty regarding the benefit of caffeine in nOH, a systematic review of its efficacy and safety was undertaken, evaluating the effectiveness and safety of caffeine on OH in adults, focussing on caffeine’s effects on BP, symptoms and adverse events.

Methods

Criteria for considering studies for this review

Participants

Adults (aged over 18 years) diagnosed with OH, as defined by the international consensus criteria in 1996 [12] or the 2011 update [1]. If diagnostic criteria were not stated, the reviewers must have been able to determine that the participants met the diagnostic criteria from the blood pressure data presented. Any underlying cause of OH was eligible for inclusion. Participants could be based in any setting (e.g. community, hospital, nursing home).

Intervention

Caffeine administered orally in any form, dose or duration. Presence of a control or comparator group was not required due to the anticipated lack of studies.

Outcomes

Studies were considered if the outcomes measured included any of the following: symptoms, diagnostic vital sign changes (e.g. orthostatic BP drop), change in resting BP or adverse effects/events.

Study type

A wide range of study types were considered in order to have a sensitive search strategy, as it was anticipated there would be a limited number of research studies on this topic. Original research studies including randomised control trials, crossover studies, observational studies and case series were eligible.

Search methods for identification of studies

Scoping work during an initial search of MEDLINE and the Centre for Reviews database (https://www.crd.york.ac.uk/CRDWeb/) was undertaken to identify keywords and terms from previous studies and review articles, to inform the search strategy. Because the scoping work did not reveal a high number of results, search terms specific to clinical trial type were not used.

Published articles were searched for using MEDLINE (1946 to week 2 January 2019), EMBASE (1974 to 22 January 2019), PubMed (no date limits) and Scopus (no date limits). Conference proceedings and theses were identified using Web of Science (1970–2019) and ProQuest (1970–2019). Grey literature was sought using Open Grey (no date limits). Ongoing or unpublished studies were searched for using the World Health Organization International Clinical Trials Registry Platform. Reviewers also searched reference lists when reviewing full-text articles. Searches were performed in January 2019. A list of search terms for each database is included in the Supplementary file.

Due to the COVID-19 pandemic, the study was paused in 2020, creating a gap between the search date and publication date. For this reason, the database search was repeated in January 2021 using the same search strategy but restricted to the dates January 2019 to January 2021. One hundred and four additional studies were identified from this update, which were all excluded in primary screening (title only).

Data collection and analysis

Selection of studies

All identified studies were collated into Endnote X9, where duplicates were removed. Primary screening was then carried out (title only), followed by secondary screening (abstract). All potentially eligible studies progressed to review of the full text to assess eligibility. All eligibility assessments were carried out by two reviewers (JG and JF).

Data extraction and management

Data were extracted from all included studies by JG and verified by JF, using forms based on the Cochrane Collaboration’s Data Collection Form for Intervention Reviews [13]. This included study design; methodology; participant characteristics; intervention nature, dosage, form of administration and duration; funding and duration of study and study outcomes. In addition to the outcomes required for inclusion, the following outcomes were extracted if they were available: activities of daily living, change in resting BP, adherence to treatment regime and orthostatic tolerance (time to onset of symptoms during upright posture).

At each stage, an independent arbitrator was available if the two reviewers disagreed; however, this was not required.

Assessment of risk of bias in included studies

The quality of the included studies was assessed independently by JG and JF. Criteria described in the Cochrane Handbook for Systematic Reviews of Interventions [13] were followed, consisting of risk of bias from selection, performance, detection, attrition, reporting and any additional bias identified. Risk of bias in each area was judged as high, low or unclear.

Data synthesis and analysis

Due to high heterogeneity in the data and incompletely reported outcomes, meta-analysis of data was not possible for any of the outcomes.

Protocol and registration

The review protocol was registered prospectively (accessed at: crd.york.ac.uk/PROSPERO/ ID: CRD42020124589). Changes to the planned protocol: Originally the study intended to study the effect of caffeine on orthostatic intolerance, including OH, postural tachycardia syndrome (PoTS) and neurally mediated syncope. However, initial scoping work revealed no relevant studies for PoTS, neurally medicated syncope or OH of non-neurogenic aetiology. Therefore, the protocol was adapted to focus solely on nOH.

Results

Study selection

The study selection process is summarised in Fig. 1.

Fig. 1
figure 1

Study selection process

Description of studies

All five included studies were based in the United States, were single-centre and were reported in English. The number of participants included in the studies ranged from 5 to 16, the mean age of participants ranged from 64 to 69 years, and all participants had neurogenic OH, which was predominantly due to Parkinson’s disease (PD), multisystem atrophy (MSA) and pure autonomic failure (PAF).

One study was a summation of case reports, and four studies were crossover trials, with participants receiving caffeine and placebo at different time points. The duration of the crossover studies ranged from 2 to 7 h, involving a single dose of oral caffeine, administered in tablet form.

The characteristics of the five studies meeting the inclusion criteria are summarised in Table 1.

Table 1 Characteristics of included studies

Participants

Participants in all five studies fulfilled international consensus criteria for OH. Three studies [14,15,16] did not state the diagnostic criteria used, but we were able to confirm OH from baseline data presented in each study.

Regarding OH severity, the mean standing BP was 86 mmHg and mean postural change in systolic blood pressure was 58.8 mmHg amongst participants from two studies (n = 24) [16, 17].

A total of 46 participants who received caffeine remained at completion. Recruitment and withdrawal data for the studies were not available.

Mean age of participants cannot be calculated with the available data, as two studies did not clarify the age of the participants involved in the arm of the studies that involved caffeine administration [15, 18].

Interventions

Caffeine was administered orally in tablet form in all five studies. In three studies the caffeine was administered in combination with ergotamine. Ergotamine is a multimodal vasoconstrictor, and caffeine has been shown to increase ergotamine’s intestinal absorption [19]. This took the form of a combination tablet in two studies [14, 17] and subcutaneous injection 30 min prior to caffeine administration in one study [15]. The caffeine dose administered ranged from 100 to 300 mg, and the mean was 189.1 ± 75.9 mg.

Methods

Three studies were randomised [15,16,17], although in two of these the method of randomisation was not specified [15, 16]. Two studies were single-blinded (participants) [17, 18]. One study was non-blinded [14], and the blinding of the remaining studies is unclear [15, 16]. All four crossover studies involved a single dose of intervention, with physiological responses being measured up to 60 min to 8 h after administration.

Effects of interventions

Findings of studies are summarised in Table 2.

Table 2 Summary of findings

Symptoms

Symptoms were reported in two studies. In Arnold’s paper [17], ergotamine/caffeine significantly reduced overall symptom severity, measured using the Orthostatic Hypotension Questionnaire’s (OHQ) [20] composite score (p = 0.034) and light-headedness component (p = 0.040) at 60 min. In contrast, there was no significant effect on symptoms with midodrine or placebo. However, the size of the effect is unclear, and each arm of the study was not compared directly.

Dewey [14] also reported symptom improvement, defined as a ‘transient or persistent reduction of symptoms during outpatient use of the drug’, in six out of eight patients who were administered ergotamine/caffeine treatment. However, the time point at which this was assessed is unclear, and there is no evidence that their method of measuring symptom burden had been validated.

Data regarding symptomatic response to caffeine as a monotherapy was not collected in any of the studies reviewed.

Orthostatic BP drop

In one study [14], when comparing baseline to post-ergotamine/caffeine (measured at 75–120 min), orthostatic SBP drop was reduced by 16.50 (± 10.11) mmHg and DBP drop was reduced by 11.33 (± 9.91) mmHg. When comparing baseline to ‘during therapy’ (unclear time point), caffeine/ergotamine treatment resulted in a reduction in orthostatic SBP drop of 44.25 (± 31.05) mmHg and DBP by 5.83 (± 19.76) mmHg.

Arnold [17] measured postural SBP 60 min post-ergotamine/caffeine, at baseline (seated) and after 1, 3, 5 and 10 min of standing. The area under the curve (AUC) was calculated and compared to placebo and midodrine. No statistical difference was found between ergotamine/caffeine and midodrine (ΔAUCSBP: −163; 95% CI −387 to 62; p = 0.155) or ergotamine/caffeine and placebo (ΔAUCSBP: 248; 95% CI −73 to 568; p = 0.130).

Change in standing blood pressure

Summation of individual participant data from Dewey’s study [14] reveals that a single dose of ergotamine/caffeine increased standing BP, with SBP rising by 40 (± 10.40) mmHg and DBP rising by 17 (± 9.17) mmHg 75–120 min after administration.

Standing SBP also increased by 42.13 (± 21.05) mmHg and DBP rose by 8.33 (± 15.19) mmHg ‘during therapy’ (time point not specified) when compared to pretreatment.

Change in seated blood pressure

Arnold’s [17] study compared the effect of ergotamine/caffeine, placebo and midodrine on seated SBP, measuring seated SBP 30 min before and 60 min post-intervention.

Ergotamine/caffeine significantly increased seated SBP compared to placebo (slope difference: 1.003; 95% CI 1.001–1.005; p = 0.003). However, in comparison to midodrine, there was no significant difference (slope difference: 1.000 95% CI 0.998–1.001; p = 0.621). Nine out of 12 participants’ seated SBP increased by  ≥ 20 mmHg with ergotamine/caffeine, compared to 5 out of 12 with midodrine, although the difference in findings was non-significant (p = 0.125), and the effect of placebo was not reported.

Summation of individual participant data from Dewey’s study [14] demonstrated that a single dose of ergotamine/caffeine increased supine/seated BP, with SBP rising by 24 (± 16.76) mmHg and DBP rising by 16 (± 17.77) mmHg on average, 75–120 min after administration. However, this effect was not seen in the longer term, with seated SBP falling by 2 (± 30.21) mmHg and DBP rising by 2 (± 12.41) mmHg ‘during therapy’ (time point not specified) when compared to pretreatment.

Hoeldtke [15] found that ergotamine/caffeine treatment increased mean arterial pressure (MAP) in five patients with OH more effectively than ergotamine or caffeine monotherapy or placebo when areas under the curve from baseline to 480 min after administration were compared (effect size not reported, p < 0.05). Graphical data from Hoeltdke’s study [15] also appears to show that caffeine monotherapy increases seated MAP compared to placebo consistently from 0 to 480 min after administration, but significance statistical testing for this was not carried out.

Onrot [16] demonstrated that administration of caffeine monotherapy lead to an initial significant rise in seated BP, from 129 ± 25/78 ± 12 at baseline to 141 ± 30/84 ± 16 mmHg after 45 min (p < 0.01). The effect on systolic blood pressure became non-significant between 75 and 90 min post-caffeine ingestion, whilst the effect of diastolic blood pressure remained significant up to 120 min (end of observation period). They also report that mean arterial pressure 1 h after caffeine ingestion was significantly higher (p < 0.05) than ‘before’ caffeine ingestion, but no data are provided to confirm this.

Jordan [18] found no significant difference in the peak seated SBP in the 120 min after administration of caffeine compared to baseline SBP or peak SBP after placebo administration within the same time period.

Orthostatic tolerance

Arnold’s study [17] reported the percentage of participants able to stand for 10 min, 1 h after administration of ergotamine/caffeine, midodrine or placebo. The result was not significantly different, with 66.6, 50 and 41.7% of participants in each arm able to stand for 10 min, respectively.

Other outcomes

Activities of daily living and adherence to therapy (in longer-term studies) were not reported in any of the identified studies.

Adverse events

The frequency and nature of adverse events were inadequately recorded (non-systematically or not at all). Hoeldtke [15] reported side effects in one out of 12 participants after single administration of caffeine tablet (heartburn). Dewey [14] reported that three out of eight participants stopped taking ergotamine/caffeine due to side effects: after 1 week due to nausea, 2 weeks due to ‘atypical chest pain’ and 14 weeks due to supine hypertension. In Arnold’s study [17], one out of five patients who continued ergotamine/caffeine after the study stopped taking the medication after an unspecified duration of time due to ‘feeling tense’. Two studies did not report adverse events [16, 18].

Risk of bias across studies

Risk of bias is summarised in Table 3. Overall, all studies were of a high risk of bias. As fewer than 10 studies were included, a funnel plot of reporting bias was precluded [13].

Table 3 Risk of bias

Discussion

This systematic review has found a lack of good-quality evidence for the use of caffeine in nOH. The studies reviewed highlight that caffeine, particularly when in combination with ergotamine, may cause short-term improvements in blood pressure and symptom burden in patients with nOH, but due to the poor quality of evidence, caffeine can only be recommended when other evidence-based treatment options have been exhausted. As no studies were identified involving participants with non-neurogenic OH, no conclusions can be drawn about caffeine’s effects in this patient group.

All included studies were small and took place in one of three sub-specialised centres. Studies were limited to participants with the alpha-synucleinopathies PD, MSA or PAF. Participants tended to be around retirement age and, based on the data presented, appear to have had relatively severe nOH. With little to no information provided about participant comorbidity, concurrent medication or performance status, it is difficult to judge how representative these participants are and whether they reflect the usual clinical patient with nOH.

In the reviewed studies, caffeine was administered as an oral tablet, either as a monotherapy or in combination with ergotamine, at a dose of 100–250 mg. There were no identified studies evaluating the effect of caffeine on OH in other preparations, such as within widely consumed hot beverages like tea or coffee.

All of the studies included in this review were found to be of high risk of bias in multiple domains, potentially a reflection of a lack of formal research reporting guidelines at the time the studies were conducted (four out of five were published between 1986 and 1998). Trial methodology, including identification, selection, randomisation, allocation and blinding, was poorly performed or poorly described. In general, inclusion and exclusion criteria and participant selection were poorly described, increasing the risk of selection bias. Attrition was also poorly addressed in all included studies. The risk of reporting bias was high in all papers. A lack of trial registration, author-derived outcomes and missing data were common, leading to potential publication bias. Furthermore, in four studies no outcome measures were provided in the methodology section [14,15,16, 18]. There were potential conflicts of interest in two studies, with one study being sponsored by the experimental drug manufacturer [15] and in the other, uncertainty over who the study sponsor was [14].

Pooled quantitative analyses could not be performed on any of the outcomes due to significant heterogeneity in the outcomes measured and incomplete reporting of data. Although all studies measured the effect of caffeine or ergotamine/caffeine on participant’s seated/supine BP, the data in four out of five papers were only displayed graphically. Only one study measured the clinically important impact on orthostatic blood pressure changes [14] and effect on symptom burden [17], and this was only after a single dose of ergotamine/caffeine. A further limitation to meta-analysis was the varied timing of outcome measurement. Although all included studies measured the very short-term effects of caffeine, the timing varied widely. The short-term nature of the outcome measurement also limits the external validity of the findings, with the effectiveness of caffeine at 1 h post-dose not being particularly clinically useful.

As meta-analysis could not be undertaken, the effect size and variance of the outcomes studied are unclear. The quality of the evidence found in this review is poor, with significant bias, in the studies reviewed. Before ergotamine/caffeine can be considered as a treatment for nOH, further larger-scale, methodologically sound studies are needed to validate the above findings. Such studies should include other clinically important outcomes, such as ability to undertake activities of daily living, falls and adverse events. These studies should also aim to evaluate the effectiveness of caffeine in the long term.

That being said, there are significant barriers to conducting the idealistic studies described above, which may also go some way to explain some of the shortcomings of the studies reviewed in this systematic review.

nOH is a rare disease requiring specialist diagnosis and management [5]. As a consequence, most experimental studies involving nOH are carried out in a limited number of sub-specialist centres with a large enough patient cohort, leading to small sample sizes and selection bias. Conducting longer-term studies into caffeine may also be challenging due to the lack of pharmaceutical funding for such a trial involving a generic drug [21]. Indeed, recent therapeutic advances in nOH have been sponsored by the pharmaceutical industry [22].

There are several limitations to this systematic review. In order to create a sensitive rather than specific search strategy, this review includes non-randomised studies, which will naturally lower the quality of evidence judgements. However, given the lack of studies in this area, this became necessary. Although a wide range of data sources were utilised in the study identification process, it did not include certain regional specific databases or non-English studies. It is possible that pooling of data could have become possible if data were sought from the study authors.

In conclusion, due to lack of good-quality evidence to support or refute its use, caffeine should only be considered as a treatment for adults with nOH when evidence-based treatments have been exhausted.