JMM

pISSN: 2288-6478, eISSN: 2288-6761 http://dx.doi.org/10.6118/jmm.2016.22.1.14 Journal of Menopausal Medicine 2016;22:14-19

Review Article

The Association between Body Mass Index and Hot Flash in Midlife Women: A Meta-analysis Fatemeh Shobeiri1, Ensiyeh Jenabi1, Jalal Poorolajal2, Seyyed Mohammad Mahdi Hazavehei3 1

Mother and Child Care Research Center, Hamadan University of Medical Sciences, Hamadan, 2Modeling of Noncommunicable Diseases Research Center, Department of Epidemiology & Biostatistics, School of Public Health, Hamadan University of Medical Sciences, Hamadan, 3 Research Center for Health Sciences, Department of Public Health, School of Public Health, Hamadan University of Medical Sciences, Hamadan, Iran

Objectives: The association between body mass index (BMI) and hot flash risk has not been specifically clarifies yet. This metaanalysis was, therefore, conducted to estimate the association between overweight and obesity and hot flash risk. Methods: We searched PubMed, Web of Science, and Scopus for observational studies addressing the association between BMI and hot flash until August 2015. Data were independently extracted and analyzed using 95% odds ratio (OR), and confidence intervals (CI) based on the random-effects models. Results: We identified 2,244 references and conducted seven studies with 4,219 participants. The association between hot flash and overweight was estimated 1.13 (95% CI: 0.97-1.32) and that of obesity was estimated 1.79 (95% CI: 1.52-2.11). No evidence of heterogeneity and publication bias was observed. Conclusion: This meta-analysis demonstrated that, though not to a great extent, obesity does increase the risk of hot flash. The findings from this meta-analysis indicated that obesity is associated with an increased risk of hot flash. Further large prospective cohort studies are required to provide convincing evidence as to whether or not BMI is associated with an increased risk of hot flashes. (J Menopausal Med 2016;22:14-19)

Key Words: Body mass index · Case-control studies · Cohort studies · Hot flashes · Meta-analysis

Introduction

The symptom of hot flashes was known to be affected by various factor such as current and former smoking, older

Approximately 40% to 70% of midlife women experience 1

hot flashes in menopausal transition periods. Hot flashes in midlife women last for a few months, around five years 1

age, hormone levels, tubal ligation, depressive symptom surgical menopausal and race.1,3~5 Some studies, however, reported that body mass

and, in certain cases, even up to 30 years. Hot flashes are

index (BMI) is the factor related to hot flashes.1 Several

characterized by the decrease in and the loss of ovarian

epidemiological studies have revealed the relationship

activity following the reduction in estrogen, permanent

between BMI and hot flash, but the results are not

2

cessation of menstruation and loss of reproductive ability.

consistent.2,6,7 Some studies have reported a positive

Such disease often causes sleep disturbance, fatigue,

association between obesity and hot flash2,6,8,9 while others

irritability, forgetfulness, acute physical discomfort and work

have not.7,10,11

problems.1

To our knowledge, no meta-analysis has addressed the

Received: October 23, 2015 Revised: November 24, 2015 Accepted: December 18, 2015 Address for Correspondence: Ensiyeh Jenabi, Mother and Child Care Research Center, Hamadan University of Medical Sciences, P.O. Box 65168-68875, Hamadan, Iran Tel: +98-81-383535019, Fax: +98-81-38380509, E-mail: [email protected] Copyright © 2016 by The Korean Society of Meno­pause This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/).

14

Fatemeh Shobeiri, et al. Association between BMI and Hot Flash

association between BMI and hot flash in midlife women,

mean/range (year), sample size, the effect measure and its

hence the present meta-analysis conducted to estimate the

95% confidence interval (CI).

overall effect of overweight and obesity on hot flash based on the current evidence.

We assessed the risk of bias among the included studies using Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) considering the following key elements: limitation, consistency, directness,

Materials and Methods

precision, reporting bias, strength, gradient, and confounding.13

1. Criteria for including studies Cohort, case-control, and cross-sectional studies were

4. Heterogeneity and publication bias

included so as to address the relationship between BMI and

The statistical heterogeneity was explored using the

hot flash, irrespective of age, race, publication date and

Q-test. The I2 statistic was used for assessing inconsistency

language. The exposure of interest was overweight and

in the results.14 The publication bias was explored employing

obesity. BMI, the body mass (kg) divided by the square of

the Egger's15 and Begg's16 tests and was visualized using a

the body height (m) is categorized into underweight (< 18.5

funnel plot.

2

2

kg/m ), normal weight (18.5-24.9 kg/m ), overweight (252

2 12

29.9 kg/m ) and obese (≥ 30 kg/m ).

The odds ratio (OR) and its associated 95% CI were mentioned as a measure of the connection between hot flash and overweight and obesity. Wherever reported, we

2. Search methods

used the full adjusted forms of OR controlled for at least

We employed a search strategy along with a set of

one or more of the potential confounding factors such as

keywords including (hot flush or hot flash) and (BMI or body

age, smoking, Hormone therapy, race, herbal supplement,

size or obese or obesity or overweight). Major electronic

tubal ligation, parity, and menopause. Data were analyzed

databases including PubMed, Web of Science, and Scopus

and the results were reported making use of a random-

were searched until August 2015. In order to find additional

effects model.17 Since the results and the number of the

references, the reference lists of the included studies were

included studies were respectively homogenous and limited,

screened. In addition, we contacted the authors of the

no subgroup analysis was performed. All statistical analyses

studies for more potentially eligible studies. Furthermore,

were performed at a significance level of 0.05 employing

the following conference databases were searched:

Stata software, version 11 (StataCorp, College Station, TX,

·The 8th European Congress on Menopause (EMAS);

USA) and Review Manager 5.

available from: www.maturitas.org/article/S03785122(09)70260-9/abstract ·The North American Menopause Society (NAMS);

Results

available from: http://www.menopause.org/annualmeetings

1. Results of the search We identified 2,244 references, 2,033 references through

3. Data collection and analysis

searching electronic database and 211 references after

Two authors independently made the decision on which

checking reference lists. No reference was found through

studies met the inclusion criteria for the objective of this

searching conference databases. One thousand two hundred

meta-analysis. Any disagreement was resolved after a

two duplicates and 1,004 irrelevant references were excluded

discussion by the authors. Two authors extracted the data

after reading titles and abstracts, while 31 were excluded after

from the included studies. The variables extracted for

reviewing full texts (Fig. 1). Seven studies remained for the

analysis contained the first author’ s name, the year and the

final meta-analysis which included 4,219 participants: two

country where the study was conducted, study design, age

cohort,4,9 two cross-sectional,1,2 and three case-control.6,8,18

http://dx.doi.org/10.6118/jmm.2016.22.1.14

15

JMM

Journal of Menopausal Medicine 2016;22:14-19

2. Effects of exposure

case-control and cohort studies, there was no significant

The relationship between overweight and hot flash is

association between hot flash and overweight (1.13; 95% CI:

shown in Fig. 2. Based on OR estimates obtained from

No of records identified through database searching (n=2,033)

0.97-1.32).

No of additional recodes identified through hand-searching (n=32)

No of records after duplicates removed (n=884)

No of records screened (n=884)

No of records excluded (n=852)

No of full text articles assessed for eligibility (n=32)

No of full-text articles excluded, with reasons (n=25)

No of studies included in quantitative synthesis (n=7)

No of studies included in quantitative synthesis (Meta-analysis) (n=7)

Study or Subgroup

Over weight Normal weight Events Total Events Total Weight

1.1.1 Case-control Gallicchio 2006 15.3% 202 139 230 124 Miller 2006 15.2% 201 135 226 124 Schilling 2007 15.7% 208 141 233 127 Subtotal (95% CI) 689 46.2% 611 Total events 415 375 2 2 2 Heterogeneity: Tau =0.00; Chi =0.05, df=2 (P=0.98); I =0% Test for overall effect: Z=0.42 (P=0.67) 1.1.2 Cohort Freeman 2001 105 175 34 61 6.6% Gallicchio 2015 168 282 128 215 17.6% Subtotal (95% CI) 457 276 24.2% Total events 273 2 162 2 2 Heterogeneity: Tau =0.00; Chi =0.24, df=1 (P=0.62); I =0% Test for overall effect: Z=0.31 (P=0.75) 1.1.3 Cross-sectional Tan 2014 9 32 18 79 2.6% Whiteman 2003 316 510 159 291 27.0% Subtotal (95% CI) 542 370 29.6% Total events 325 2 177 2 2 Heterogeneity: Tau =0.00; Chi =0.00, df=1 (P=0.97); I =0% Test for overall effect: Z=2.11 (P=0.03) Total (95% CI) 1,610 1,335 100.0% Total events 973 2 754 2 2 Heterogeneity: Tau =0.00; Chi =2.50, df=6 (P=0.87); I =0% Test for overall effect: Z=1.59 (P=0.11) 2 2 Test for subgroup differences: Chi =2.21, df=2 (P=0.33); I =9.4%

Odds ratio M-H, Random, 95% CI

Odds ratio M-H, Random, 95% CI

1.04 [0.71, 1.53] 1.09 [0.74, 1.60] 1.02 [0.70, 1.50] 1.05 [0.84, 1.31]

1.19 [0.66, 2.15] 1.00 [0.70, 1.44] 1.05 [0.77, 1.43]

1.33 [0.52, 3.37] 1.35 [1.01, 1.81] 1.35 [1.02, 1.78]

1.13 [0.97, 1.32] 0.1

0.2

0.5

1

2

5

10

Favours [normal weight] Favours [over weight]

Fig. 2. Forest plot of the association between hot flash and overweight. CI: confidence interval.

16 http://dx.doi.org/10.6118/jmm.2016.22.1.14

Fig. 1. Flow of information through the different phases of the metaanalysis.

Fatemeh Shobeiri, et al. Association between BMI and Hot Flash

The connection between obesity and hot flash is illustrated

The quality of the studies assessed using GRADE. The five

in Fig. 3. As can be seen, unlike overweight, a significant

studies were moderate in quality while the two other studies

association exists between hot flash and obesity (1.79; 95%

had low qualities (Table S1 in supplement files).

CI: 1.52-2.11).

Discussion

3. Heterogeneity and publication bias The I2 statistics and Cochran homogeneity test revealed no evidence of heterogeneity among the included studies

The results of this meta-analysis indicated that

addressing the association between hot flash and overweight

overweight and obesity may increase the risk of hot flashes.

(0%, P = 0.87) and obesity (10%, P = 0.29).

One finding in this review was an apparent dose-response

Publication bias was explored using Begg's and Egger's

relationship between BMI and a hot flash where the risk of

tests and was visualized employing a funnel plot. The Egger

a hot flash was higher among obese women than overweight

regression test and Begg's test revealed no evidence of

women; the higher the BMI, the higher the risk of hot flash

publication bias among the included studies that addressed

is. it should be noted that a dose-response relationship

addressing the association between hot flash and overweight

is strong evidence for a causal relationship.19 Although no

(P = 0.911 and P = 0.138) and obesity (P = 0.877 and P =

heterogeneity was observed in the included studies, owing to

0.621), respectively. No evidence of publication bias was

the limited number of studies there was little evidence as to

observed in the funnel plots (Fig. S1, S2 in supplement files)

the association between BMI and hot flash.

either.

Study or Subgroup

The results of this meta-analysis indicated that obesity

Obese Normal weight Events Total Events Total Weight

1.2.1 Case-control 14.9% 139 261 180 Gallicchio 2006 124 14.9% 135 261 180 Miller 2006 124 15.3% 141 270 185 Schilling 2007 127 45.0% 792 545 Subtotal (95% CI) 415 Total events 375 2 2 2 Heterogeneity: Tau =0.00; Chi =0.05, df=2 (P=0.98); I =0% Test for overall effect: Z=5.96 (P