Fermented milk for hypertension

Abstract

Background

Fermented milk has been suggested to have a blood pressure lowering effect through increased content of proteins and peptides produced during the bacterial fermentation. Hypertension is one of the major risk factors for cardiovascular disease world wide and new blood pressure reducing lifestyle interventions, such as fermented milk, would be of great importance.

Objectives

To investigate whether fermented milk or similar products produced by lactobacilli fermentation of milk proteins has any blood pressure lowering effect in humans when compared to no treatment or placebo.

Search methods

The Cochrane Central Register of Controlled Trials (CENTRAL), English language databases, including MEDLINE (1966—2011), EMBASE (1974—2011), Cochrane Complementary Medicine Trials Register, Allied and Complementary Medicine (AMED) (1985—2011), Food science and technology abstracts (1969—2011).

Selection criteria

Randomised controlled trials; cross over and parallel studies evaluating the effect on blood pressure of fermented milk in humans with an intervention period of 4 weeks or longer.

Data collection and analysis

Data was extracted individually by two authors, afterwards agreement had to be obtained before imputation in the review.

Main results

A modest overall effect of fermented milk on SBP was found (MD ‐2.45; 95% CI ‐4.30 to ‐0.60), no effect was evident on DBP (MD ‐0.67; 95% CI ‐1.48, 0.14).

Authors' conclusions

The review does not support an effect of fermented milk on blood pressure. Despite the positive effect on SBP the authors conclude, for several reasons, that fermented milk has no effect on blood pressure. The effect found was very modest and only on SBP, the included studies were very heterogeneous and several with weak methodology. Finally, sensitivity and subgroup analyses could not reproduce the antihypertensive effect. The results do not give notion to the use of fermented milk as treatment for hypertension or as a lifestyle intervention for pre‐hypertension nor would it influence population blood pressure.

Plain language summary

Fermented milk for hypertension

High blood pressure is a major health problem as it is one of the main risk factors for cardiovascular disease. Fermented milk has been suggested to reduce blood pressure in humans. This review was conducted to establish whether the intake of fermented milk lowers blood pressure in humans. In 15 studies including 1232 participants a very modest lowering of systolic blood pressure was found, but no effect on diastolic blood pressure was present. The included studies were of a variable quality and the findings do not support the use of fermented milk as antihypertensive treatment or as a lifestyle intervention to reduce blood pressure.

Background

Description of the condition

Hypertension is a common condition and a strong risk factor for cardiovascular morbidity and mortality. Target organ damage is quite often present; especially sensitive are blood vessels, heart, kidneys and brain. In 1999 the World Health Organisation estimated that hypertension causes 7.1 million deaths per year, which is 13% of total deaths worldwide (WHO 1999). In a meta‐analysis of 61 prospective studies of blood pressure (BP) and mortality, a 2 mmHg decrease of systolic blood pressure (SBP) reduced cardiac mortality by 7% and stroke mortality by 10%. The correlation between reduced mortality and BP reduction was still evident for individuals usually considered normotensive, with BP as low as 115/75 (Lewington 2002). Pre‐hypertension, SBP = 130–139 mmHg and/or diastolic blood pressure (DBP) = 80–89 mmHg, is also related to an enhanced risk of cardiovascular disease, and 50% of pre‐hypertensive individuals develop hypertension within 4 years (Julius 2006). Lifestyle modifications alone or in combination with medical treatment of hypertension are recommended to reduce BP and the risk of cardiovascular disease (ESH 2003).

Description of the intervention

Different lifestyle modifications, i.e. exercise, weight reduction, dietary changes and salt restriction, have been shown to lower BP (Dickinson 2006). Fermented milk has been suggested to reduce BP in humans and therefore, as a lifestyle modification, a BP lowering effect of fermented milk deserves careful consideration. Fermented milk is currently on the market as a functional food with putative antihypertensive properties in the USA, Spain, UK, Finland, Switzerland, Italy, South Korea, Japan, Iceland and Portugal (Boelsma 2009).

How the intervention might work

During fermentation lactic acid bacteria cell‐wall proteinases hydrolyses milk proteins, hereby increasing the amount of bioactive peptides in the fermented product. The concentration of different peptides is highly dependent on the bacteria strain used for fermentation (Chantal 2003). The research on fermented milk and BP has mainly focused on lactotripeptides (LTP) with in vitro ACE‐inhibitor effect. However, fermented milk products contain several proteins, peptides and minerals, all with possible different antihypertensive modes of actions. In addition to the presumed ACE‐inhibitor effect, reduced sympathetic activity and an opioid agonist effect have been suggested. Studies using spontaneously hypertensive rats fed fermented milk are persuasive, repeatedly finding reductions of SBP (Fuglsang 2002, Sipola 2002).

Why it is important to do this review

In contrast to the evidence from animal studies, the results of interventional human studies of the antihypertensive effect of fermented milk are diverging, with SBP changes ranging from 0 to 12 mmHg (Usinger 2009). The presumed ACE‐inhibitor effect has not been shown in man. The bioavailability and in vivo actions in humans of the peptides in fermented milk are widely unknown. As outlined, the putative antihypertensive effect in humans is not unambiguous and collecting the existing evidence will hopefully help determine if fermented milk has a role in the treatment of hypertension or as a lifestyle intervention in a population strategy.

Objectives

The aim of this review is to examine whether fermented milk or similar products produced by lactobacilli fermentation of milk proteins has BP lowering effect in humans when compared to no treatment or placebo.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials.

Types of participants

Male or female, age 18 or above.

Types of interventions

Products produced by fermentation of milk proteins by lactic acid bacteria.

Oral administration at any dose or any frequency.

Minimum four weeks duration.

Control: placebo or no treatment.

Types of outcome measures

Primary outcomes

Change of SBP or DBP from baseline compared to no treatment or placebo.

If different measurements of BP have been made, results are used in the following order of priority:

1. office BP

2. 24 hour ambulatory BP (average of all measurements)

3. home measurements

   1. sitting

   2. supine

   3. standing

Secondary outcomes

None.

Search methods for identification of studies

Electronic searches

The Database of Abstracts of Reviews of Effectiveness (DARE) and the Cochrane Database of Systematic Reviews were searched for related reviews.

The following electronic databases were searched for primary studies:

Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE (1950—2011), EMBASE (1974—2011), Cochrane Complementary Medicine Trials Register, AGRICOLA (1970‐2011), International Pharmaceutical Abstracts (IPA) (1970‐2011), and Food Science and Technology Abstracts (FSTA) (1969—2011).

MEDLINE (Appendix 1) was searched using a standard Cochrane search strategy. This search strategy was adapted to EMBASE (Appendix 2), CENTRAL (Appendix 3), AGRICOLA (Appendix 4), IPA (Appendix 5), FSTA (Appendix 6) and the other databases.

Searching other resources

The following additional resources were searched:

1. International Clinical Trials Registry Platform (WHO ICTRP)

2. Reference lists of all eligible studies and relevant reviews identified

3. Authors of relevant papers were contacted regarding any further published or unpublished work    

4. Authors of trials reporting incomplete information were contacted to provide the missing information 

5. ISI Web of Science will be searched for papers which cite studies included in the review     

6. OpenSIGLE (System for Information on Grey Literature in Europe) 

Data collection and analysis

Selection of studies

Studies which were clearly irrelevant were excluded from the titles alone. Titles and abstracts were searched for studies meeting the inclusion criteria independently by two authors (LU, CR). Full text manuscripts were retrieved for papers of all potentially eligible studies and read independently by the two authors; LU with knowledge of the area and CR, who is not a content expert. All manuscripts were independently reviewed following an established trial selection form (Appendix 7). Where disagreement occurred, the inclusion of the study was discussed. All studies, where agreement on inclusion was reached were included. If doubt of the eligibility of a study arose, the study investigators were contacted for further information. If disagreement between the two authors persisted after all information was available, a third author (HI) would decide whether the actual study should be included or not. The authors had full study reports without concealment of any information during assessment of eligibility.

Data extraction and management

Two authors (LU, CR) extracted data independently filling a predefined data extraction table. The extracted data was compared and if disagreement occurred it was solved through discussion. If disagreement persisted, a third author (HI) would decide which data to extract. In the case of missing data the principal investigators of the study were contacted. Extracted data included:

• Number of participants screened, randomised and completing the study.

• Number of intervention groups.

• Patients characteristics: age, gender, BP, antihypertensive medication (yes/no).

• Description of placebo product.

• Description of intervention; strain and species of lactic acid bacteria used, carrier, dose, frequency and length of intervention.

• Baseline SBP/DBP and SBP/DBP at the end of intervention.

• Methods used to measure BP.

Data on study methodology, results and assessment of bias is shown in the included or excluded studies table in the review.

Assessment of risk of bias in included studies

The risk of bias in each study was estimated independently by two authors and a table for assessing risk of bias was made for each study. "The risk of bias" table Figure 1 evaluate sequence generation, allocation concealment, blinding, subjects lost to follow up, selective outcome reporting and incomplete outcome data for each study as recommended in the Cochrane Handbook (Higgins 2008). After agreement between the authors, data was analysed using the Cochrane Review Manager software, RevMan 5. A graph summarizing the risk of bias of all studies was generated and included in the review.

1.

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

Measures of treatment effect

Treatment effect was determined as change in SBP/DBP from baseline in the fermented milk group and the placebo at the end of the intervention. If different doses were given the data from the group receiving the highest dose was used.

When BP was measured with several measures in the same study, data was extracted according to the following priority: office BP, 24‐hour ambulatory BP (average of all measurements), home BP measurements (1. sitting, 2. supine, 3. standing).

Unit of analysis issues

Cross over trials were included, but as the presumed antihypertensive mechanism of fermented milk in humans is widely unknown, a carry‐over effect cannot be excluded. Therefore, in the included cross over trials, only data from the first intervention period was used in the analysis in this review.

Dealing with missing data

Standard deviation for each group was found in the manuscripts or calculated from available data whenever possible. When standard deviation for each group could not be calculated, study investigators were contacted (by email, letter or fax). If the standard deviation was impossible to obtain or calculate, the average standard deviation from the other studies was used.

In case of insufficient data the study was excluded.

Assessment of heterogeneity

Standard chi‐square statistics was used to test for heterogeneity between studies. The fixed effect model was used to analyse pooled data but as the test for heterogeneity was statistically significant P<0.01, the random effect model was applied.

Data synthesis

Data synthesis and analysis were done using the Cochrane Review Manager Software, RevMan 5.

Weighted mean difference and 95% confidence interval between treatment and placebo or no treatment was calculated.

Subgroup analysis and investigation of heterogeneity

Subgroup analysis was planned according to:

1. Inclusion BP: normotensive (office SBP<140 and DBP<90) versus hypertensive (office SBP≥140 or DBP≥90). The studies were classified according to the average baseline SBP.

2. Length of intervention period: 4—8 weeks versus more than 8 weeks.

3. Studies including participants using antihypertensive medication versus studies excluding participants using antihypertensive medication.

4. According to the method used to measure BP: 24 hour ambulatory BP versus home measurements versus office BP.

Sensitivity analysis

Sensitivity analysis was planned according to:

1. High quality versus low quality studies as estimated from the Cochrane risk of bias tool (Higgins 2008).

2. Industrial sponsored versus non industrial sponsored.

Results

Description of studies

Results of the search

The result of the search yielded 1234 studies. From abstracts and titles we identified 36 studies, from which we obtained full manuscripts for further examination. 24 studies fulfilled the inclusion criteria. We included studies with milk or casein fermented with several bacteria strands ‐ not only lactobacilli as originally stated in the protocol. Studies with peptides from casein prepared by enzyme hydrolysis with bacterial cell wall proteases were included as the peptides in the product are similar to the peptides in fermented milk. Out of the 24 studies fulfilling the inclusion criteria 9 were excluded due to missing data (Figure 2). We tried to obtain the missing data from the authors by contacting them by email or letter before exclusion.

2.

Study flow diagram.

Included studies

Design

Fifteen studies were included in this review. The studies were all placebo‐controlled, four studies used a cross over design (Cicero 2010; Tuomilehto 2004; van Mierlo 2009; van der Zander 2008a).

The length of the studies varied from 4 to 23 weeks, with intervention periods of 4‐21 weeks. In six studies periods of 1‐4 weeks were applied (Cicero 2010; Engberink 2008; Germino 2010; Jauhiainen 2005; Mizuno 2005; Seppo 2003). Most studies were double‐blind, but three were single‐blind (Inoue 2003; Kawase 2000; Mizuno 2005). One study consisted of two different arms where one received a fermented milk with added potassium and the other arm a similar fermented milk without added potassium (van Mierlo 2009). Another study had separate intervention groups according to the participants' baseline BP (Mizuno 2005). Engberink et al. had three different intervention groups: LTP produced synthetic, enzymatic hydrolysis and bacterial fermentation (Engberink 2008). Finally, two studies had several intervention groups with different concentration levels of the active product (Mizuno 2005; Usinger 2010).

All studies, except Cicero 2010, had industry funding and/or some or all of the authors were employed by private food companies. Similary in all these studies the test and placebo products were produced by the involved companies.

The studies were conducted from 1996 and onwards.

Intervention

The interventions in the 15 included studies were very heterogeneous. Fermented milk was used in eight studies (Inoue 2003; Hata 1996; Jauhiainen 2005; Kawase 2000; Mizushima 2004; Seppo 2003; Tuomilehto 2004; Usinger 2010), freeze dried LTP after bacterial fermentation in four studies (Cicero 2010; Engberink 2008; Germino 2010; Mizuno 2005), and freeze dried LTP made by enzymatic processing with bacteria cell wall proteases in three studies (Sano 2005; van der Zander 2008a; van Mierlo 2009). Freeze dried peptides were distributed as tablets or with a carrier, either skim milk, yogurt or juice. For details on bacteria strain and dose see Characteristics of included studies.

Participants

A total of 1232 participants (650 males, 582 females) were included in this review. Average number of subjects included was 88, ranging from 20 to 275. Age ranged from 18 to 86. Inclusion office BP ranged from SBP 129‐180 mmHg and DBP 84‐110 mmHg, in eight studies hypertensive subjects with office BP above 140 mmHg systolic or 90 mmHg diastolic were included (Engberink 2008; Germino 2010; Inoue 2003; Hata 1996; Jauhiainen 2005; Seppo 2003; Tuomilehto 2004; van der Zander 2008a), in six of the remaining studies subjects with office SBP above 129 mmHg were eligible (Cicero 2010; Mizuno 2005; Mizushima 2004; Sano 2005; Usinger 2010; van Mierlo 2009). Kawase et al. had no lower limit of BP as inclusion criteria and one study additionally used 24hBP (SBP 130‐160 mmHg) as an inclusion criteria (van Mierlo 2009). Antihypertensive medications were allowed in two studies, as long as dose and type did not change during the study period (Hata 1996; Seppo 2003).

Outcomes

SBP and DBP was the primary outcome in all studies, except one (Kawase 2000). Either office BP; (Engberink 2008; Germino 2010; Inoue 2003; Hata 1996; Mizuno 2005; Mizushima 2004; Sano 2005; Tuomilehto 2004; van der Zander 2008a) home BP (Seppo 2003) or 24hBP (Cicero 2010; Jauhiainen 2005; Usinger 2010; van Mierlo 2009). In the study of Kawase et al. serum lipid level was the primary outcome. In this review office BP was primarily analysed as stated in the protocol. This was impossible in the study of Seppo et al. where only results from home measurements were reported and therefore used in the analysis.

Excluded studies

Most of the references identified by the first search were excluded because they were clearly irrelevant. The most common reason for exclusion was an outcome other than BP or interventions different from fermented milk. From the 36 manuscripts assessed by two reviewers, six studies were excluded because the intervention was not fermented milk or peptides from milk or casein processed with bacterial fermentation or bacteria proteases (Cadee 2007; Charlton 2008; Chin‐Dusting 2006; Lee 2007;Sacks 1987; Townsend 2004). One double publication was identified (Jauhiainen 2007a) and five studies appeared not to be randomised (Ashar 2004; Hirata 2002; Kajimoto 2001; Kajimoto2002; Nakamura 2004). Finally, nine studies met the inclusion criteria, but had to be excluded due to insufficient data, which could not be obtained despite attempts made to contact the study authors (Aihara 2005; de Leeuw 2009; Ishida 2006; Ishida 2007; Itakura 2001; Kajimoto 2004gerholm‐Larsen 2000; van der Zander 2008; Yoshizawa 2009).

Risk of bias in included studies

Many of the included studies were lacking information on the methodology (Figure 1).

Allocation

Most of the studies lacked information concerning sequence generation and allocation sequence concealment. The randomisation process was only clearly described in two studies (Cicero 2010; Engberink 2008), additional information on randomisation was retrieved from four studies by successful contact with the authors (Inoue 2003; Kawase 2000; Usinger 2010; van Mierlo 2009). All authors were contacted for further information but the remaining did not respond. The last eight studies were reported as randomised studies, but information of the exact methodology of randomisation was sparse.

Blinding

The blinding of both participants and employees was well described in eight studies and three studies were single‐blind (Inoue 2003; Kawase 2000; Mizuno 2005). In the remaining four studies all described as double‐blind, descriptions on how blinding was implemented were not available (Germino 2010; Sano 2005; Hata 1996; van der Zander 2008a).

Incomplete outcome data

In only one study did the authors fail to account for withdrawals (Mizuno 2005). Attrition in the included studies was generally low; the cumulative total number of withdrawals for all studies was 40 participants.

Selective reporting

As no protocols were available we could not examine whether there was selective reporting, but in all included studies differences in BP were reported.

Other potential sources of bias

All studies, except Cicero 2010, were supported by industrial sponsors with financial interests in the products and in several studies some of the authors were employees in the companies producing the test products. This is known to cause publication bias (Higgins 2008).

Effects of interventions

An overall effect was found on office SBP (MD ‐2.45; 95% CI ‐4.30, ‐0.60), but no effect on DBP (MD ‐0.67; 95% CI ‐1.48, 0.14) (Analysis 1.1; Analysis 1.5). According to our protocol four subgroup analyses were performed:

1.1. Analysis.

Comparison 1 Fermented milk versus placebo, Outcome 1 Systolic blood pressure.

1.5. Analysis.

Comparison 1 Fermented milk versus placebo, Outcome 5 Diastolic blood pressure.

• The length of the intervention (more than 8 weeks compared to 4‐8 weeks) did not influence the effect on BP (Analysis 1.2, Analysis 1.6).

• Analyses of subgroups according to baseline BP: in the studies including subjects with SBP above 140 mmHg no effect on BP was present: Considering the studies including participants with baseline SBP below 140 mmHg an effect was seen separately on SBP (Analysis 1.3, Analysis 1.7).

• The subgroup including studies with participants taking antihypertensive medication yielded 2 studies (Hata 1996; Seppo 2003) in which an effect was shown on both SBP and DBP. In the subgroup without antihypertensive medication the effect remained, although reduced, on SBP, but continuously no effect was seen on DBP (Analysis 1.4, Analysis 1.8)

• Analysis of office BP were carried out as data from our primary outcome; office BP was available in all included studies except one (Seppo 2003). The analysis of the studies with 24hBP data showed no effect on SBP (MD ‐0.80; 95% CI ‐2.72 to 1.12), nor DBP (MD ‐0.74; 95% CI ‐1.67 to 0.20).

1.2. Analysis.

Comparison 1 Fermented milk versus placebo, Outcome 2 Length of intervention period SBP.

1.6. Analysis.

Comparison 1 Fermented milk versus placebo, Outcome 6 Length of intervention group DBP.

1.3. Analysis.

Comparison 1 Fermented milk versus placebo, Outcome 3 Baseline SBP high versus low SBP.

1.7. Analysis.

Comparison 1 Fermented milk versus placebo, Outcome 7 Baseline SBP high versus low DBP.

1.4. Analysis.

Comparison 1 Fermented milk versus placebo, Outcome 4 Antihypertensive medication SBP.

1.8. Analysis.

Comparison 1 Fermented milk versus placebo, Outcome 8 Anthypertensive medication DBP.

Sensitivity analysis were to be performed in two areas:

• Industry funding versus no industry funding; this sensitivity analysis was not conducted because all studies, except Cicero 2010, were funded by the industry.

• High quality versus low quality studies. After removing the eight studies with high risk of bias (Germino 2010; Inoue 2003; Hata 1996; Kawase 2000; Mizuno 2005; Sano 2005; Tuomilehto 2004; van der Zander 2008a) no significant effects were seen as a result of the intervention on neither SBP (MD ‐0.57 mmHg; 95% CI ‐3.12 to 1.98) nor DBP (MD ‐0.39 mmHg; 95% CI ‐1.78 to 1.01). High risk of bias was defined as more than two unclear or one high risk of bias results in the risk of bias tables (Figure 1).

Discussion

Summary of main results

Despite a modest main effect on SBP, the authors remain sceptical that fermented milk or similar products produced by lactobacilli fermentation of milk proteins has any BP lowering effect in humans when compared to no treatment or placebo. Sensitivity analysis without the studies with imputed SD (Inoue 2003; Mizuno 2005; van Mierlo 2009) could only show a very modest significant effect on SBP (MD ‐2.07 mmHg; 95% CI ‐4.04, ‐0.11) and none on DBP (MD ‐0.90 mmHg; 95% ‐1.90, 0.10). Furthermore the included studies were of variable quality and when excluding the studies with a high risk of bias no significant effect were found on either SBP nor DBP.

The main challenge while conducting this review was to compare such very different interventions: tablets with freeze dried LTP versus milk fermented with different bacteria strains or fermented milk with enzymatically produced LTP. Since the presumed antihypertensive mechanism of fermented milk is unclear this heterogeneity could influence the results. A previous meta analysis including only studies with the two most commonly used LTP (VPP and IPP) found an effect on both SBP (4.8 mmHg; 95% CI 3.7– 6.0) and DBP (2.2 mmHg; 95% CI 1.3–3.1)(Xu 2008), however this meta analysis by Xu et al. did not include the latest larger studies of LTP, all showing no effect or a very modest effect. (Cicero 2010; Engberink 2008; Germino 2010; van der Zander 2008a; Usinger 2010; de Leeuw 2009; van Mierlo 2009; van der Zander 2008).

All subgroup analyses, except two, showed no effect on BP. In the subgroup analysis of the studies which included participants who took antihypertensive medication an antihypertensive effect of fermented milk was seen. This subgroup analysis only yielded two studies (Hata 1996; Seppo 2003). In the group with low baseline SBP an effect of fermented milk was found, this is unexplained and probably by pure chance.

The most accurate method for measurement of BP is considered to be 24hBP (Neutel 1999; Parati 2002; Stergiou 2002). 24hBP was reported in five of the included studies (Engberink 2008; Cicero 2010Jauhiainen 2005; Usinger 2010; van Mierlo 2009). Meta analysis of the available data from these studies showed no effect on SBP (MD ‐0.80; 95% CI ‐2.72 to 1.12) or DBP (MD ‐0.74; 95% CI ‐1.67 to 0.20). Besides these studies 24hBP have been measured in two intervention studies, excluded due to insufficient data (de Leeuw 2009; van der Zander 2008), but in which likewise negative results were reported.

The ACE‐inhibition of fermented milk, shown in animal‐ and in vitro experiments, has not been confirmed in man. In five of the studies in this review ACE‐activity, plasma concentrations of angiotension I and II or aldosterone have been measured to evaluate the presumed ACE‐inhibition of fermented milk, but all the results were insignificant (Engberink 2008; Cicero 2010; Jauhiainen 2005; Mizushima 2004; Usinger 2010).

Overall completeness and applicability of evidence

The participants in the studies of this review represented a typical pre‐hypertensive and hypertensive population. The studies were conducted in Asia and Europe, but the results should be applicable to any population world wide.

Quality of the evidence

The quality of the studies varied significantly. There was a tendency of a higher methodological quality in the more recently published studies with negative results. This further supports the negative finding of this review.

Unfortunately as many as nine studies had to be excluded due to missing data, which weakens the results of this review considerable. Hopefully these data can be retrieved and included in the next update of this review. Data in these studies were reported unclear in the manuscripts, without reporting the BP differences from baseline.

Potential biases in the review process

We estimate that the potential bias in this review is low. Despite all efforts to follow the protocol we made two assumptions. We extended the inclusion criteria, after we saw the diversity of test products in the available studies. The objectivity in this process during the work with the review is not assessable. We furthermore made minor changes in the subgroup analyses from what was originally intended, but this is not likely to have any impact on the results. Two of the authors (LU and HI) of this review have conducted an RCT on fermented milk and BP with negative results (Usinger 2010) and this could have potentially affected their overall objectivity.

Agreements and disagreements with other studies or reviews

To date only two meta analysis have been conducted, both including studies with the two most commonly used LTP (VPP and IPP). As mentioned, an effect on both SBP and DBP was found by Xu et al. (Xu 2008). In the latest review by Cicero et al. a positive effect on BP by LTP was shown, in similarity with our results the effect is most pronounced in Asian studies. This is possibly due to a weaker methodology or as speculated by Cicero et al. a stronger effect in Asians for unknown reasons (Cicero 2011). Several narrative reviews elaborating on the antihypertensive effect of fermented milk are published with different conclusions (Boelsma 2009; Geleijnse 2010; Jauhiainen 2007; Usinger 2009; Takano 2002).

Authors' conclusions

Implications for practice.

The results of this review do not support the use of fermented milk as treatment for hypertension or as lifestyle intervention for pre‐hypertension. There is insufficient evidence from randomized controlled trials to suggest that fermented milk has a BP lowering effect.

Implications for research.

It is not likely that further studies will yield different results, but studies with higher content and /or better bioavailability of the active peptides are warranted.

Appendices

Appendix 1. MEDLINE search strategy

Database: Ovid MEDLINE(R) 1948 to Present with Daily Update 
 Search Date: 8 July 2011 
 Search Strategy: 
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 
 1 milk/ (38254) 
 2 exp milk proteins/ (27332) 
 3 exp cultured milk products/ (1465) 
 4 exp lactobacillaceae/ (16344) 
 5 (yogourt or yogurt or yoghurt or kefir or buttermilk or lactobacill$ or lactococc$ or leuconost$ or casein or lactotripeptid$).tw. (37998) 
 6 lactic acid bacter$.tw. (4302) 
 7 ((milk or dairy or cheese$ or cream$) adj2 (ferment$ or cultur$ or product$ or sour$ or protein$)).tw. (17612) 
 8 or/1‐7 (102124) 
 9 randomized controlled trial.pt. (311115) 
 10 controlled clinical trial.pt. (82828) 
 11 randomized.tw. (232192) 
 12 placebo$.tw. (130712) 
 13 randomly.tw. (157464) 
 14 trial.tw. (269654) 
 15 groups.tw. (1062236) 
 16 or/9‐15 (1573982) 
 17 animals/ not (humans/ and animals/) (3533167) 
 18 16 not 17 (1279579) 
 19 exp hypertension/ (190496) 
 20 (pre‐hypertens$ or hypertens$).tw. (264700) 
 21 anti‐hypertensive agents/ (45727) 
 22 (antihypertensive$ or anti‐hypertensive$ or lactotripeptid$).tw. (34077) 
 23 (blood pressure or bloodpressure).mp. (316292) 
 24 (diastolic or systolic or arterial).tw. (361272) 
 25 heart rate/ (128598) 
 26 cardiovascular agents/ (7140) 
 27 (heart or cardiovascular or cardioprotective).tw. (657662) 
 28 (angiotensin or ace or inhibit$).tw. (1437367) 
 29 or/19‐28 (2518775) 
 30 8 and 18 and 29 (846)

Appendix 2. EMBASE search strategy

Database: EMBASE (Ovid) <1980 to 2011 8 July > 
 Search Strategy: 
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 
 1 exp dairy product/ (57059) 
 2 exp lactobacillaceae/ (21284) 
 3 fermented product/ (821) 
 4 exp lactic acid bacterium/ (2884) 
 5 lactic acid bacter$.tw. (5157) 
 6 (yogourt or yogurt or yoghurt or kefir or zabadi or buttermilk or lactobacill$ or lactococc$ or leuconost$ or casein or lactotripeptid$).tw. (42002) 
 7 ((milk or dairy or cheese$ or cream$) adj2 (ferment$ or cultur$ or product$ or sour$ or protein$)).tw. (19657) 
 8 or/1‐7 (106839) 
 9 randomized controlled trial/ (282041) 
 10 crossover procedure/ (29993) 
 11 double‐blind procedure/ (99051) 
 12 random$.tw. (631995) 
 13 (crossover$ or cross‐over$).tw. (55219) 
 14 placebo$.tw. (155954) 
 15 (doubl$ adj blind$).tw. (115850) 
 16 assign$.tw. (178267) 
 17 allocat$.tw. (59255) 
 18 or/9‐17 (930461) 
 19 (animal$ not (human$ and animal$)).mp. (3276690) 
 20 18 not 19 (835207) 
 21 exp hypertension/ (360687) 
 22 (pre‐hypertens$ or hypertens$).tw. (322828) 
 23 anti‐hypertensive agent/ (52562) 
 24 (antihypertensive$ or anti‐hypertensive$ or lactotripeptid$).tw. (43796) 
 25 (blood pressure or bloodpressure).mp. (340682) 
 26 (diastolic or systolic or arterial).tw. (414975) 
 27 heart rate/ (123763) 
 28 cardiovascular agent/ (7091) 
 29 (heart or cardiovascular or cardioprotective).tw. (784265) 
 30 (angiotensin or ace or inhibit$).tw. (1586328) 
 31 or/21‐30 (2894409) 
 32 8 and 20 and 31 (747)

Appendix 3. Cochrane Central search strategy

Database: EBM Reviews ‐ Cochrane Central Register of Controlled Trials (Ovid) <8 July 2011> 
 Search Strategy: 
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 
 #1 MeSH descriptor Milk, this term only 773 
 #2 MeSH descriptor Milk Proteins explode all trees 607 
 #3 MeSH descriptor Cultured Milk Products explode all trees 227 
 #4 MeSH descriptor Lactobacillaceae explode all trees 800 
 #5 (yogourt or yogurt or yoghurt or kefir or buttermilk or lactobacill* or lactococc* or leuconost* or casein or lactotripeptid*):ti,ab in Clinical Trials 1613 
 #6 lactic acid bacter*:ti,ab in Clinical Trials 123 
 #7 (milk or dairy or cheese* or cream*) near/2 (ferment* or cultur* or product* or sour* or protein*):ti,ab in Clinical Trials 818 
 #8 (#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7) 3129 
 #9 MeSH descriptor Hypertension explode all trees 12546 
 #10 (pre‐hypertens* or hypertens*):ti,ab in Clinical Trials 22001 
 #11 MeSH descriptor Antihypertensive Agents explode all trees 20621 
 #12 (antihypertensive* or anti‐hypertensive* or lactotripeptid*):ti,ab in Clinical Trials 6973 
 #13 (blood pressure or bloodpressure):ti,ab in Clinical Trials 32800 
 #14 (diastolic or systolic or arterial):ti,ab in Clinical Trials 33032 
 #15 MeSH descriptor Heart Rate, this term only 14486 
 #16 MeSH descriptor Cardiovascular Agents explode all trees 56808 
 #17 (heart or cardiovascular or cardioprotective):ti,ab in Clinical Trials 46668 
 #18 (angiotensin or ace or inhibit*):ti,ab in Clinical Trials 38285 
 #19 (#9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18) 136685 
 #20 (#8 AND #19) 390

Appendix 4. AGRICOLA search strategy

Database: AGRICOLA (Ovid) <1970 to November 2009> 
 Search Strategy: 
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 
 1 exp dairy product/ (23146) 
 2 exp lactobacillaceae/ (5655) 
 3 lactic acid bacter$.tw. (4008) 
 4 (yogourt or yogurt or yoghurt or kefir or zabadi or buttermilk or lactobacill$ or lactococc$ or leuconost$ or casein or lactotripeptid$).tw. (18925) 
 5 ((milk or dairy or cheese$ or cream$) adj2 (ferment$ or cultur$ or product$ or sour$ or protein$)).tw. (21605) 
 6 or/1‐5 (57371) 
 7 random$.tw. (31571) 
 8 placebo$.tw. (3243) 
 9 (crossover$ or cross‐over$).tw. (2967) 
 10 trial.tw. (14528) 
 11 assign$.tw. (14539) 
 12 allocat$.tw. (9056) 
 13 (doubl$ adj blind$).tw. (2277) 
 14 or/7‐13 (62937) 
 15 (human$ not (human$ and animal$)).tw. (71239) 
 16 14 not 15 (60031) 
 17 exp hypertension/ (2201) 
 18 (pre‐hypertens$ or hypertens$).tw. (3968) 
 19 (antihypertensive$ or anti‐hypertensive$ or lactotripeptid$).tw. (523) 
 20 (blood pressure or bloodpressure).mp. (5779) 
 21 (diastolic or systolic or arterial).tw. (4042) 
 22 heart rate/ (2120) 
 23 cardiovascular agents/ (183) 
 24 (heart or cardiovascular or cardioprotective).tw. (19884) 
 25 (angiotensin or ace or inhibit$).tw. (106234) 
 26 or/17‐25 (132104) 
 27 6 and 16 and 26 (238)

Appendix 5. IPA search strategy

Database: International Pharmaceutical Abstracts (Ovid) <1970 to June 2011> 
 Search Strategy: 
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 
 Search Strategy: 
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 
 1 milk$.mp. (1341) 
 2 (yogourt or yogurt or yoghurt or kefir or buttermilk or lactobacill$ or lactococc$ or leuconost$ or casein or lactotripeptid$).mp. (466) 
 3 lactic acid bacter$.mp. (21) 
 4 ((dairy or cheese$ or cream$) adj2 (ferment$ or cultur$ or product$ or sour$ or protein$)).mp. (112) 
 5 or/1‐4 (1849) 
 6 random$.mp. (34456) 
 7 placebo$.tw. (19544) 
 8 (crossover$ or cross‐over$).tw. (7778) 
 9 trial.tw. (22452) 
 10 assign$.tw. (7147) 
 11 allocat$.tw. (1664) 
 12 (doubl$ adj blind$).tw. (16313) 
 13 or/6‐12 (56898) 
 14 (pre‐hypertens$ or hypertens$).tw. (12285) 
 15 (antihypertensive$ or anti‐hypertensive$ or lactotripeptid$).tw. (4054) 
 16 (blood pressure or bloodpressure).tw. (9203) 
 17 (diastolic or systolic or arterial).tw. (7621) 
 18 (heart or cardiovascular or cardioprotective).tw. (20205) 
 19 (angiotensin or ace or inhibit$).tw. (39694) 
 20 or/14‐19 (67797) 
 21 5 and 13 and 20 (28)

Appendix 6. FSTA search strategy

Database: Food Science and Technology Abstracts (Ovid) <1969 to 2011 June> 
 Search Strategy: 
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ 
 1 milk$.mp. (95891) 
 2 (yogourt or yogurt or yoghurt or kefir or buttermilk or lactobacill$ or lactococc$ or leuconost$ or casein or lactotripeptid$).mp. (40895) 
 3 lactic acid bacter$.mp. (10549) 
 4 ((dairy or cheese$ or cream$) adj2 (ferment$ or cultur$ or product$ or sour$ or protein$)).mp. (45511) 
 5 or/1‐4 (134920) 
 6 random$.mp. (13475) 
 7 placebo$.tw. (1649) 
 8 (crossover$ or cross‐over$).tw. (1730) 
 9 trial.tw. (7889) 
 10 assign$.tw. (5436) 
 11 allocat$.tw. (1630) 
 12 (doubl$ adj blind$).tw. (1392) 
 13 or/6‐12 (24889) 
 14 (pre‐hypertens$ or hypertens$).tw. (1550) 
 15 (antihypertensive$ or anti‐hypertensive$ or lactotripeptid$).tw. (809) 
 16 (blood pressure or bloodpressure).tw. (1847) 
 17 (diastolic or systolic or arterial).tw. (968) 
 18 (heart or cardiovascular or cardioprotective).tw. (10030) 
 19 (angiotensin or ace or inhibit$).tw. (52028) 
 20 or/14‐19 (62688) 
 21 5 and 13 and 20 (377)

Appendix 7. Trial selection form

Study ID	Report ID	Citation and contact details	Review author ID and date
	yes	unclear	no	notes
1. Is the study randomised?
2. Age of the participants (above 18)?
3. Is the intervention product milk proteins adapted with lactic acid bacteria?
4. Is the product given orally, minimum once daily?
5. Is the intervention period 4 weeks or more?
6. Is the BP measured at baseline and at the end of intervention?
Final decision

Data and analyses

Comparison 1. Fermented milk versus placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Systolic blood pressure	15	1232	Mean Difference (IV, Random, 95% CI)	‐2.45 [‐4.30, ‐0.60]
2 Length of intervention period SBP	15	1232	Mean Difference (IV, Random, 95% CI)	‐2.45 [‐4.30, ‐0.60]
2.1 intervention 4‐8 weeks	9	780	Mean Difference (IV, Random, 95% CI)	‐2.26 [‐4.37, ‐0.15]
2.2 Intervention longer than 8 weeks	6	452	Mean Difference (IV, Random, 95% CI)	‐3.02 [‐7.16, 1.13]
3 Baseline SBP high versus low SBP	15	1232	Mean Difference (IV, Random, 95% CI)	‐2.45 [‐4.30, ‐0.60]
3.1 Baseline SBP above 140	11	881	Mean Difference (IV, Random, 95% CI)	‐2.31 [‐5.10, 0.47]
3.2 Baseline SBP below 140	5	351	Mean Difference (IV, Random, 95% CI)	‐3.28 [‐4.51, ‐2.04]
4 Antihypertensive medication SBP	15	1232	Mean Difference (IV, Random, 95% CI)	‐2.45 [‐4.30, ‐0.60]
4.1 No antihypertensive medication allowed	13	1163	Mean Difference (IV, Random, 95% CI)	‐2.08 [‐3.98, ‐0.18]
4.2 Antihypertensive medication allowed	2	69	Mean Difference (IV, Random, 95% CI)	‐7.37 [‐13.06, ‐1.69]
5 Diastolic blood pressure	15	1232	Mean Difference (IV, Random, 95% CI)	‐0.67 [‐1.48, 0.14]
6 Length of intervention group DBP	15	1232	Mean Difference (IV, Random, 95% CI)	‐0.67 [‐1.48, 0.14]
6.1 Intervention 4‐8 weeks	9	780	Mean Difference (IV, Random, 95% CI)	‐0.66 [‐1.60, 0.28]
6.2 Intervention longer than 8 weeks	6	452	Mean Difference (IV, Random, 95% CI)	‐0.56 [‐2.35, 1.23]
7 Baseline SBP high versus low DBP	15	1232	Mean Difference (IV, Random, 95% CI)	‐0.67 [‐1.48, 0.14]
7.1 Baseline SBP above 140 DBP	11	881	Mean Difference (IV, Random, 95% CI)	‐0.31 [‐1.26, 0.64]
7.2 Baseline SBP below 140 DBP	5	351	Mean Difference (IV, Random, 95% CI)	‐1.42 [‐2.40, ‐0.44]
8 Anthypertensive medication DBP	15	1232	Mean Difference (IV, Random, 95% CI)	‐0.67 [‐1.48, 0.14]
8.1 No antihypertensive medication allowed	13	1163	Mean Difference (IV, Random, 95% CI)	‐0.50 [‐1.30, 0.29]
8.2 Antihypertensive medication allowed	2	69	Mean Difference (IV, Random, 95% CI)	‐3.98 [‐7.37, ‐0.60]

Comparison 2. 24hBP.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 SBP	6		Mean Difference (Random, 95% CI)	‐0.80 [‐2.72, 1.12]
2 DBP	5		Mean Difference (Random, 95% CI)	‐0.74 [‐1.67, 0.20]

2.1. Analysis.

Comparison 2 24hBP, Outcome 1 SBP.

2.2. Analysis.

Comparison 2 24hBP, Outcome 2 DBP.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Cicero 2010.

Methods	Randomised, placebo‐controlled, double‐blind, crossover study
Participants	Normal (SBP 120‐129/DBP 80‐84) or high‐normal (SBP 130‐139/DBP 85‐90) blood pressure, no antihypertensive medication
Interventions	Fruit juice added 3 mg VPP/IPP twice a day
Outcomes	24hBP and Office BP
Bacteria	No information
Notes	No interest conflicts
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation done by letting the participants drawing envelopes with codes unknown by the investigators
Allocation concealment (selection bias)	Low risk	Randomisation done by letting the participants drawing envelopes with codes unknown by the investigators
Blinding (performance bias and detection bias) Participants	Low risk	Bottles from a blinded box
Blinding (performance bias and detection bias) Investigators	Low risk	Bottles from a blinded box
Incomplete outcome data (attrition bias) Office BP	Unclear risk	5 withdrawals, all explained
Selective reporting (reporting bias)	Low risk	none identified
Other bias	Unclear risk	Low compliance only 65% of the test products consumed.

Engberink 2008.

Methods	Randomised, placebo‐controlled, double‐blind study
Participants	458 screened, 135 randomised. Offfice SBP above 140 mmHg, no antihypertensive medication. Age 35‐70
Interventions	Low fat yogurt 200 ml once daily. Intervention product: Added LTP concentrated from fermented milk. Placebo: 200 ml low fat yogurt without LTP added. Intervention period 8 weeks.
Outcomes	Office BP, home BP and 24hBP
Bacteria	No information
Notes	Four intervention groups; Synthetic produced LTP, LTP from FM, Enzymatic produced LTP, Placebo. The group with LTP from FM and placebo group are included in the data analysis. Office BP was measured in all participants, 24hBP only in 58 out of 135, data from office BP is used in the analysis. Sponsored by Unilever Research and Developement Center Vlaardingen. The sponsor provided test products and financial support
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation codes
Allocation concealment (selection bias)	Low risk	Two independent persons not involved in the conduct of the study
Blinding (performance bias and detection bias) Participants	Low risk	The test drinks were similar in taste, colour and served in non‐transparent cups.
Blinding (performance bias and detection bias) Investigators	Low risk	The staff was blinded during the study and data‐analysis
Incomplete outcome data (attrition bias) Office BP	Low risk	Office BP was measured at all participants, except 1 withdrawal. 58 of 135 participants accepted 24hBP measurements. Analysis of 46; 12 had less than 70% readings or below 20 hours measure
Selective reporting (reporting bias)	Low risk	None identified
Other bias	Low risk	None identified

Germino 2010.

Methods	Randomised, placebo‐controlled, double‐blind study
Participants	123 screened, 91 included. Stage 1 and 2 Hypertensives, no antihypertensive medication.
Interventions	VPP and IPP derived from milk protein casein in powder form dispensed in juice
Outcomes	24hBP and Office BP
Bacteria	No information
Notes	All authors have study grants or are advisors for Calpis USA Inc.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	All patients were randomised in a 2:1 ratio, randomisation procedure not further described
Allocation concealment (selection bias)	Unclear risk	All patients were randomised in a 2:1 ratio, randomisation procedure not further described
Blinding (performance bias and detection bias) Participants	Unclear risk	Not described
Blinding (performance bias and detection bias) Investigators	Unclear risk	Not described
Incomplete outcome data (attrition bias) Office BP	Low risk	10 withdrawals, all explained
Selective reporting (reporting bias)	Unclear risk	Endpoint in the study was daytime 24hBP and effect in "dippers", the only statistically significant results, protocol sought but could not be found.
Other bias	Unclear risk	None identified, low compliance (65% of the distributed products consumed)

Hata 1996.

Methods	Randomised, placebo‐controlled, double‐blind study
Participants	36 hypertensive subjects, (hypertension not defined) included, 32 of the participants were receiving antihypertensive medication.
Interventions	Fermented milk 100 ml x1. Intervention product: Skim milk fermentation 37º, 22 hours then pasturized. Placebo product: Skim milk acidified with 2,3% lactic acid. Intervention period 8 weeks.
Outcomes	Office BP
Bacteria	L. Helveticus, S. Cerevisiae
Notes	Fermented milk products from Calpis Food Industri and cooperation on the study with Calpis.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Subjects were randomly assigned, no further information
Allocation concealment (selection bias)	Unclear risk	Subjects were randomly assigned, no further information
Blinding (performance bias and detection bias) Participants	Unclear risk	Not described
Blinding (performance bias and detection bias) Investigators	Unclear risk	The nurse measuring BP was blinded to the treatment assignment
Incomplete outcome data (attrition bias) Office BP	Low risk	All drop outs were explained
Selective reporting (reporting bias)	Low risk	None Identified
Other bias	Unclear risk	Screening and inclusion criteria are not described

Inoue 2003.

Methods	Randomised, placebo‐controlled, single‐blind study
Participants	39 mildly hypertensive subjects, (SBP 140‐159 mmHg and DBP 90‐99 mmHg) included, antihypertensive medication not allowed. Age 28‐81.
Interventions	Fermented milk 100 ml x1. Intervention product: Skim milk fermentation with two strains. Placebo product: Skim milk acidified with 2,3% lactic acid. Intervention period 8 weeks. Presumed active peptide; GABA.
Outcomes	Office BP
Bacteria	Lb. Casei (Shirota), Lc. lactis YIT 2027
Notes	Fermented milk products from Yakult Honsha Co. LTD. Several authors from Yakult Central Institute. Data provided by author.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	The doctor randomised the hypertensives to active or placebo group in order of arrival at the hospital.
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding (performance bias and detection bias) Participants	Low risk	Described as single‐blind
Blinding (performance bias and detection bias) Investigators	High risk	Single‐blind
Incomplete outcome data (attrition bias) Office BP	Low risk	4 withdrawals, all explained
Selective reporting (reporting bias)	Low risk	None identified
Other bias	Low risk	None identified

Jauhiainen 2005.

Methods	Randomised, placebo‐controlled, double‐blind study.
Participants	139 participants screened, 108 included. 94 analysed with 24hBP. Inclusion criteria SBP 140‐180 and DBP 90‐110, no antihypertensive medication allowed.
Interventions	Fermented milk 150 ml x2, Intervention period 10 weeks.
Outcomes	24hBP and office BP
Bacteria	Intervention product fermented with L. Helveticus LBK‐16H, placebo with Lactococcus sp. culture
Notes	Test products made by Valio Ltd. Study made in cooperation with Valio Ltd.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomly allocated
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding (performance bias and detection bias) Participants	Low risk	All test products in coded blank cartons
Blinding (performance bias and detection bias) Investigators	Low risk	All test products in coded blank cartons
Incomplete outcome data (attrition bias) Office BP	Low risk	Clearly described
Selective reporting (reporting bias)	Low risk	None identified
Other bias	Low risk	None identified

Kawase 2000.

Methods	Randomised, placebo‐controlled, single‐blind
Participants	20 healthy men with cholesterol above 200 mg/L, no antihypertensive medication allowed. BP is not a part of the inclusion criteria. Only males, age 30‐51.
Interventions	200 ml Fermented milk x1. Bacteria in the end intervention product. Placebo acidified with lactic acid. Intervention period 8 weeks.
Outcomes	Office BP
Bacteria	L Casei TMC0409, S Thermophilus TMC1543
Notes	Primary Outcome Cholestol. Data provided by author
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Simple randomisation
Allocation concealment (selection bias)	Unclear risk	Simple randomisation
Blinding (performance bias and detection bias) Participants	Low risk	described as single blind
Blinding (performance bias and detection bias) Investigators	High risk	single‐blind
Incomplete outcome data (attrition bias) Office BP	Low risk	No withdrawals
Selective reporting (reporting bias)	Low risk	None identified
Other bias	Unclear risk	Baseline BP are lower in the placebo group

Mizuno 2005.

Methods	Randomised placebo‐controlled, single‐blind study
Participants	136 screened, 131 included. SBP 120‐159 and DBP 85‐99 Divided in 2 groups: high normal blood pressure (SBP 120‐139) and mild hypertensive (SBP 140‐159). Age 30‐57, no antihypertensive medication allowed.
Interventions	Intervention product: Casein hydrolysis of A. Oryzae protease, spray‐dried and distributed in tablets; 1,8/2,5/3,6 mg, once daily, 6 weeks intervention. Placebo: 0 mg, added sodium caseinate to reach same concentration of amino nitrogen.
Outcomes	Office BP
Bacteria	A. Oryzae
Notes	The study had four groups: Placebo, 1,8 mg, 2,5 mg and 3,6 mg tablets, we analysed placebo and 3,6 mg Test products from Calpis and several authors employees of Calpis Food Research Laboratory
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomised. Subjects were divided into 4 groups without any difference of age, BP, hearth rate and weight.
Allocation concealment (selection bias)	Unclear risk	Randomly allocated
Blinding (performance bias and detection bias) Participants	Low risk	Tablets with unknown concentration of LTP
Blinding (performance bias and detection bias) Investigators	High risk	Described as single blind
Incomplete outcome data (attrition bias) Office BP	High risk	No withdrawals
Selective reporting (reporting bias)	Low risk	None identified
Other bias	Low risk	None identified

Mizushima 2004.

Methods	Randomised placebo‐controlled double‐blind
Participants	46 men with office SBP above 129 and DBP above 84, age 23‐59. No antihypertensive medication allowed.
Interventions	160 gr x1 fermented milk or placebo daily for 4 weeks
Outcomes	Office BP
Bacteria	L. Helveticus and S. Cerevisiae. Placebo product artificially acidified
Notes	One author from Calpis Food Industri CO, Ltd.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	subjects were randomly and blindly allocated to two groups
Allocation concealment (selection bias)	Low risk	subjects were randomly and blindly allocated to two groups
Blinding (performance bias and detection bias) Participants	Low risk	Did not know which group they were assigned
Blinding (performance bias and detection bias) Investigators	Low risk	subjects were blindly allocated to one of two groups. BP was measured by two investigators who were blinded to treatment assignments
Incomplete outcome data (attrition bias) Office BP	Low risk	4 dropouts all explained
Selective reporting (reporting bias)	Low risk	None identified
Other bias	Low risk	None identified

Sano 2005.

Methods	Randomised, placebo controlled, double‐blind study
Participants	150 included, 144 completed the study. SBP 130‐159 and/or DBP 85‐99, no antihypertensive medication allowed. Age above 20.
Interventions	200 ml x1, 12 weeks. Test product: casein hydrolysate made with protease from A. Oryzae, juice as carrier. Placebo without LTP
Outcomes	Office BP
Bacteria	A. Oryzae
Notes	Study products was produced by Calpis CO, Ltd, Made in cooperation with Calpis CO, Ltd.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described
Allocation concealment (selection bias)	Unclear risk	subjects were divided into groups by a controller who did not participate in the study in any other way
Blinding (performance bias and detection bias) Participants	Unclear risk	double‐blind study, but no further description
Blinding (performance bias and detection bias) Investigators	Unclear risk	double‐blind study, but no further description
Incomplete outcome data (attrition bias) Office BP	Low risk	6 withdrawals, all explained
Selective reporting (reporting bias)	Low risk	None identified
Other bias	Low risk	None identified

Seppo 2003.

Methods	Randomised, placebo‐controlled, double blind study
Participants	45 screened, 42 included, 39 completed the study. SBP above 140 or DBP above 90, antihypertensive medication allowed. Age 30‐51.
Interventions	Fermented milk 150 ml x1, 21 weeks. Intervention product: Milk fermented with L Helveticus LBK‐16H. Placebo: Mesofilic Lactococcus sp.
Outcomes	Home BP measurements
Bacteria	Intervention product: L Helveticus LBK‐16H Placebo: Mesofilic Lactococcus sp
Notes	The test products was produced by Valio Ltd. Made in cooperation with Valio Ltd.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described
Allocation concealment (selection bias)	Unclear risk	Seperate randomisation list according to the use of antihypertensive medication.
Blinding (performance bias and detection bias) Participants	Low risk	The placebo and intervention product was delivered in blank cartons and the participants did not know which group they were allocated to.
Blinding (performance bias and detection bias) Investigators	Low risk	Staff was blinded
Incomplete outcome data (attrition bias) Office BP	Low risk	All withdrawals explained, 3 before the study, another 3 did not complete the study.
Selective reporting (reporting bias)	Low risk	None identified
Other bias	Low risk	None identified

Tuomilehto 2004.

Methods	Randomised, placebo‐controlled, double‐blind, cross over study
Participants	60 participants. SBP 140‐180 and/or DBP 90‐110. No antihypertensive medication allowed. Mean age 52 years.
Interventions	Fermented milk 150 ml X1, first intervention period:10 weeks followed by 3‐4 weeks wash out, second intervention period of 5‐7 weeks. Only data from the first intervention period are used in this review
Outcomes	Office BP
Bacteria	Intervention product: L. Helveticus LBK‐16.
Notes	Test products and funding from Valio Ltd.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	subjects were randomised not further described
Allocation concealment (selection bias)	Unclear risk	Not described
Blinding (performance bias and detection bias) Participants	Low risk	Products and groups were coded.
Blinding (performance bias and detection bias) Investigators	Low risk	Staff were blinded
Incomplete outcome data (attrition bias) Office BP	Low risk	60 subjects were randomised, 2 withdrawals in the first intervention period. Data from 1 person was used in the analysis, as withdrawal was in week 9.
Selective reporting (reporting bias)	Low risk	None identified
Other bias	High risk	The study design was converted to a cross over design in the end of first intervention period due to missing significant results because of an unexpected strong placebo effect.

Usinger 2010.

Methods	Randomised, placebo controlled, double‐blind study
Participants	SBP 130‐149 or DBP 85‐95, no antihypertensive medication allowed. 258 screened, 94 included, 90 completed the study
Interventions	Fermented milk 150 or 300 ml x1. Intervention period 8 weeks. Intervention product fermented with L. Helveticus. Placebo product was acidified with glucono‐δ‐lactone 1,75%
Outcomes	24hBP, office BP
Bacteria	L. Helveticus Cardi04
Notes	Funding and test products provided by Chr. Hansen A/S. In the study there were two intervention groups 300 ml and 150 ml, only the group receiving 300 ml are included in this review.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer generated sequence of numbers, coincidently allocated number to any subject, randomised in blocks of 12
Allocation concealment (selection bias)	Low risk	Treatment group was unknown to participants and staff at the time of randomisation.
Blinding (performance bias and detection bias) Participants	Low risk	Intervention and placebo product in similar looking and coded bottles
Blinding (performance bias and detection bias) Investigators	Low risk	Bottles coded, staff remained blinded until after analysis of BP data.
Incomplete outcome data (attrition bias) Office BP	Low risk	4 withdrawals, all explained
Selective reporting (reporting bias)	Low risk	None identified
Other bias	Unclear risk	Baseline BP are significantly lower in the placebo group

van der Zander 2008a.

Methods	Randomised, placebo‐controlled, double‐blind, cross over study
Participants	SBP 130‐160 and DBP below 100, no antihypertensive medication allowed. 1097 screened, 275 included, 271 completed.
Interventions	Fermented milk 250 ml x1 daily, 4 weeks intervention 4 weeks washout, 4 weeks intervention. Test product: fermented milk with LTP added. Placebo: Milk fermented with the same bacteria but without LTP added.
Outcomes	Office BP
Bacteria	L. Helveticus and Saccharomyces cerevisiae
Notes	The authors are employees of Unilever, who produced the test products and funded the study
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomly assigned to receive either test or the control product.
Allocation concealment (selection bias)	Unclear risk	Randomly assigned to receive either test or the control product.
Blinding (performance bias and detection bias) Participants	Unclear risk	double‐blind, no further description
Blinding (performance bias and detection bias) Investigators	Unclear risk	double‐blind, no further description
Incomplete outcome data (attrition bias) Office BP	Low risk	4 withdrawals, all explained.
Selective reporting (reporting bias)	Low risk	None identified
Other bias	Low risk	None identified

van Mierlo 2009.

Methods	Randomised, placebo‐controlled, double‐blind, cross over study. Two multicenter studies. Study 1:LTP, study 2:LTP and potassium
Participants	24hBP: SBP 130‐160 or office BP above 135 and DBP below 100. No antihypertensive medication allowed. 263 screened, 162 randomised. Study 1: 69, study 2: 93.
Interventions	Yogurt drink daily for 4 weeks; 4 weeks wash out between the intervention periods. Study 1:Test product: yogurt drink added freeze dried enzymatically produced LTP. Placebo: yogurt drink added whey protein without LTP. 200 ml x1. Study 2 Test product: yogurt drink added freeze dried enzymatically produced LTP and potassium. Placebo: yogurt drink added whey protein without LTP. 100 ml x1
Outcomes	24hBP and office BP
Bacteria	Both drinks acidified with a standard yogurt strain. LTP produces by enzymatic hydrolyses and added to the test products.
Notes	All authors were working for Unilever Food and Health research. Unilever provided test products
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomly assigned according in two blocks according to baseline BP (letter from author)
Allocation concealment (selection bias)	Unclear risk	Randomly assigned
Blinding (performance bias and detection bias) Participants	Low risk	All products in nontransparent coded cups
Blinding (performance bias and detection bias) Investigators	Low risk	Staff remained blinded during the whole study period and data analysis.
Incomplete outcome data (attrition bias) Office BP	Low risk	All withdrawals are explained
Selective reporting (reporting bias)	Low risk	None identified
Other bias	Low risk	None identified

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Aihara 2005	Insufficient data
Ashar 2004	Not randomised: Information from the author; first 20 persons were given the test product, the next 10 the placebo product
de Leeuw 2009	Insufficient data
gerholm‐Larsen 2000	Randomisation unclear, Insufficient data.
Hirata 2002	Not randomised
Ishida 2006	Insufficient data
Ishida 2007	Insufficient data
Itakura 2001	Insufficient data
Jauhiainen 2007a	Double publication, with focus on arterial stiffness
Kajimoto 2001	Not randomised
Kajimoto 2004	Insufficient data
Kajimoto2002	Not randomised
Nakamura 2004	Not randomised
van der Zander 2008	Insufficient data
Yoshizawa 2009	Insufficient data

Differences between protocol and review

We expanded the inclusion criteria from milk fermented with lactobacillus to milk fermented with all types of bacteria as well as products with LTP produced by bacterial fermentation or hydrolysis with bacteria proteases of casein or milk proteins.

Contributions of authors

LU formulated the idea for the review, wrote the protocol, extracted data and wrote the first version of the review. CR extracted data and edited the review. CR and HI have read several drafts of the protocol and the review and contributed to the final version.

Sources of support

Internal sources

• University Hospital of Glostrup, Denmark.

External sources

• No sources of support supplied

Declarations of interest

LU and HI have conducted a RCT of fermented milk and are the authors of a narrative review of fermented milk.

LU worked at department of clinical physiology and nuclear medicine, Glostrup Hospital 2006‐2008, the department received financial support to conduct a RCT concerning fermented milk from Chr. Hansen A/S, Denmark.

HI has given lectures at scientific meetings organised by pharmaceutical companies for more than 35 years, has been member of advisory boards and member of steering commitees for studies supported by pharmaceutical companies.

New