Obesity poses a major public health problem worldwide, driving an epidemic of associated complications such as type 2 diabetes mellitus and CVD(Reference James1). People of sub-Saharan African ancestry (blacks) in Western countries exhibit higher prevalences of obesity and obesity-related cardiovascular disease risk factors than people of European ancestry (whites).
Surveys in the USA have indicated that the prevalence of obesity is 32·2 % in the general population, 30·6 % in whites and 45·0 % in blacks. Among men, the prevalence is similar in whites (31·1 %) and blacks (34·0 %). However, black women are more likely to be obese (53·9 %) than white women (30·2 %)(Reference Ogden, Carrol and Curtin2). In addition to higher obesity rates, blacks have higher prevalences of CVD risk factors, including higher prevalences of diabetes, hypertension and stroke(Reference Wild and McKeigue3–Reference Chaturvedi, McKeigue and Marmot6), and higher mortality rates from CVD(7) compared with whites.
The higher prevalence of obesity and CVD risk factors in blacks can be ameliorated by effective weight management. However, this is both a clinical and a public health challenge, as evidence from trials indicates that blacks underachieve in weight management programmes compared with whites(Reference Parikh, Lo and Chang8–Reference McMahon, Fujioka and Singh13). Various reasons have been proposed for this including social and cultural barriers such as differential body-image ideals, cultural food attitudes, fewer models for physical activity, and normative views of overweight and obesity(Reference Cooper, Kennelly and Durazo-Arvizu14, Reference Kumanyika15).
No review has been identified focusing on the effect of weight change by dietary and other lifestyle changes on risk factors in blacks. The present review is therefore opportune, given the higher cardiovascular risk in blacks, exacerbated by limited access to the knowledge or resources required to incorporate dietary and lifestyle changes(Reference Kumanyika16).
Methods
Data sources and study selection
The following sources were searched: MEDLINE, EMBASE, CINAHL and CCTR (Cochrane Controlled Trials Register). Reference lists of original studies and other systematic reviews were also examined. Keywords used included ‘blacks’ or ‘black Africans’ or ‘African Americans’ or ‘Afro-Caribbeans’ or ‘black British’ or ‘black Americans’ and ‘obesity’ or ‘overweight’ or ‘diabetes’ or ‘heart disease’ or ‘cardiovascular disease’ or ‘hypertension’ in combination with ‘intervention’ or ‘trial’ using various suffixes. Electronic searches and reviewing of results were performed by one reviewer (G.O.A.).
Another reviewer (C.B.) assessed the relevance of identified studies for inclusion, and disagreements were resolved by consensus. Studies were included on the basis of: (i) dietary intervention with/without lifestyle change (behaviour change and physical activity) v. control; (ii) randomised or non-randomised controlled trials; (iii) black participants (for studies made up of mixed ethnicity, authors were contacted for subgroup results on blacks); (iv) duration of intervention ≥3 months; (v) studies published between January 1990 and December 2009; and (vi) reported weight/BMI change and change in at least one of the following: waist circumference, systolic and diastolic blood pressures, fasting plasma lipids and glucose, and glycated haemoglobin (HbA1c).
Data synthesis
A QUOROM (quality of reporting of meta-analysis) statement was used to describe how studies identified through the searches were processed(Reference Moher, Cook and Eastwoods17). Each study was summarized with regard to characteristics of participants and interventions, duration and dropout rate. Because of the various dietary interventions employed, results of included studies were not pooled but rather expressed as the average mean difference between intervention and control using StatsDirect Statistical Software version 2·7·7 (StatsDirect Ltd, Altrincham, UK).
Results
Findings and description of studies
Search results are summarized in the QUOROM flow diagram (Fig. 1). Eighteen studies met the inclusion criteria out of thirty-five potential studies(Reference Agurs-Collins, Kumanyika and Have18–Reference McNabb, Quinn and Kerver35). Seven of the studies recruited healthy obese participants, ten enrolled only participants with diabetes(Reference Agurs-Collins, Kumanyika and Have18, Reference Anderson-Loftin, Barnett and Bunn19, Reference Racette, Weiss and Obert21, Reference Samuel-Hodge, Keyserling and Park23, Reference West, Dilillo and Bursac26, Reference Mayer-Davis, D'Antonio and Smith30, Reference Ziemer, Berkowitz and Panayioto31, Reference Barnard, Cohen and Jenkins33–Reference McNabb, Quinn and Kerver35) and one study recruited participants with hypertension(Reference Elmer, Obarzanek and Vollme32). Intervention duration ranged from 3·5 to 12 months (median of 6 months).
Fig. 1 QUOROM (quality of reporting of meta-analysis) flowchart
All studies took place in the USA, and about 50 % of the included trials recruited less than 100 participants in the either intervention or control arm. Furthermore all studies employed dietary interventions/advice with lifestyle modifications (behaviour change and physical activity).
The majority (fifteen of eighteen) of the trials that were described as randomised specified eligibility criteria, but none explicitly stated the method of randomisation and details of allocation concealment in the trials. Incomplete outcome data were addressed in ten of eighteen trials. Six studies recruited only women participants(Reference Resnicow, Taylor and Baskin22, Reference Yanek, Becker and Moy24, Reference West, Dilillo and Bursac26–Reference Kennedy, Champagne and Ryan29) and the rest were of mixed gender. Most of the studies were clinic- or community-based, and only three were church-based(Reference Resnicow, Taylor and Baskin22–Reference Yanek, Becker and Moy24). Attrition rate ranged from 6 % to 32 %. Table 1 summarises the characteristics of the participants and interventions.
Table 1 Key characteristics of studies included in the present review on dietary interventions for weight loss and cardiovascular risk reduction in blacks
Outcomes
All studies showed a positive treatment effect on weight change between intervention and control except two studies(Reference Becker, Yanek and Johnson20, Reference Samuel-Hodge, Keyserling and Park23). Net average weight loss ranged from 5·40 kg in 3·5 months(Reference McNabb, Quinn and Kerver35) to 2·49 kg in 6 months(Reference Agurs-Collins, Kumanyika and Have18, Reference Anderson-Loftin, Barnett and Bunn19, Reference Resnicow, Taylor and Baskin22, Reference West, Dilillo and Bursac26–Reference Kennedy, Champagne and Ryan29, Reference Ziemer, Berkowitz and Panayioto31–Reference Barnard, Cohen and Jenkins33) and 2·91 kg in 12 months(Reference Becker, Yanek and Johnson20, Reference Racette, Weiss and Obert21, Reference Samuel-Hodge, Keyserling and Park23–Reference Zemel, Richards and Milstead25, Reference Mayer-Davis, D'Antonio and Smith30, 34). Average mean difference in weight loss was −2·66 kg for all studies, −2·63 kg for studies that recruited healthy participants and −2·76 kg for studies that enrolled only patients with type 2 diabetes. Even though weight loss in type 2 diabetes patients was higher, it was not significantly different from weight loss in all studies or in healthy participants (Table 2).
Table 2 Outcome measures (expressed as average mean difference between intervention and control) of dietary interventions for weight loss and cardiovascular risk reduction in blacks(Reference Agurs-Collins, Kumanyika and Have18–Reference McNabb, Quinn and Kerver35)
Weight loss was associated with net improvements in waist circumference (−2·95 cm), fasting blood glucose (−0·82 mmol/l), HbA1c (−0·51 %), systolic (−1·4 mmHg) and diastolic blood pressures (−0·6 mmHg), LDL cholesterol (−0·06 mmol/l), HDL cholesterol (+0·31 mmol/l) and TAG (−0·29 mmol/l), but not in total cholesterol. No significant difference was detected between all studies and the studies of healthy participants or type 2 diabetes patients for any outcome (Table 2).
Discussion
The higher prevalence of cardiovascular risk factors in blacks has been attributed to multiple influences of genetics, socio-economic factors and lifestyle that promote obesity and make weight loss difficult(Reference Chiu, Province and Permutt36–Reference Tull and Roseman40). Reducing CVD risk factors in blacks by weight control has therefore not been effective, probably due to a lack of cultural competency in weight management programmes coupled with a failure to address issues of relevance to the population(Reference Kumanyika and Ewart41, Reference Maillet, Melkus and Spollett42). Additionally there is paucity of high-quality data on diet and lifestyle interventions for weight and cardiovascular risk reduction in people of African ancestry, and this gap presents a major hurdle in developing effective policies and programmes(Reference Yancey, Kumanyika and Ponce43).
Reviews on weight management interventions in minorities have been published recently(Reference Seo and Sa44, Reference Osei-Assibey, Kyrou and Adi45), but the magnitude of the average net treatment effect on cardiovascular risk reduction has not been studied in people of African ancestry. Eighteen out of thirty-five potential studies fulfilled the inclusion criteria of the present review. Most of the potential studies that were excluded targeted weight management without assessment of its impact on cardiovascular risk reduction.
The net average weight loss of 2·7 kg found in the present review is lower than that reported previously in mostly white Caucasian populations (−5·0 kg, range −0·40 to −7·80 kg) using standard 2510 kJ/d (600 kcal/d) deficit and low-fat diets for ≥12 months(46). However, the results are comparable to analyses which showed, for each kilogram of weight loss, that systolic and diastolic blood pressures fall by 1·0 mmHg each(Reference Neter, Stam and Kok47) and blood parameters improve: total cholesterol by −0·13 mmol/l, TAG by −0·09 mmol/l, LDL cholesterol by −0·05 mmol/l, HDL cholesterol by +0·004 mmol/l(Reference Dattilo and Kris-Etherton48) and fasting plasma glucose by −0·2 mmol/l(Reference Anderson and Konz49).
Although the clinical significance of the results of the present review cannot be fully determined, one cannot dismiss its public health relevance in blacks. This is because studies have intimated that reductions in the incidence of type 2 diabetes and CVD can be achieved by sustained lifestyle changes(Reference Agurs-Collins, Kumanyika and Have18, 50–52).
Results of the Diabetes Prevention Programme (DPP) in which 19·9 % of participants were blacks(50, Reference West, Prewitt and Bursac51) and the Look AHEAD study(34) in which 15·5 % of participants were blacks all provide a standard for weight management for the control of risk factors in blacks. Findings of the DPP and Look AHEAD should therefore be of interest to clinicians and public health researchers who are interested in implementing interventions in people of African ancestry. It must however be recognised that participants in the DPP were rigorously selected to be highly motivated and the intervention was resource-intensive to deliver(Reference Samaha, Iqbal and Seshadri28, Reference West, Prewitt and Bursac51).
Some of the studies included in the present review suffered from limitations, including small number of participants, lack of power to detect significant changes and high attrition rates. Some of the studies attempted to address the higher attrition rates by using the last observation carried forward (LOCF). However, such analysis underestimates or overestimates the results, depending on different dropout rates in treatment and control groups(Reference Padwal, Li and Lau53).
The present review has highlighted a serious deficiency of published research in an increasingly important area in people of African ancestry. Although interesting, the studies that have been published have used various approaches which have suggested improvement in weight and cardiovascular risk factor reduction in the medium term (6 months). However, changing attitudes may not by themselves lead to sustained behaviour change if the environment is not supportive of these changes(Reference Kumanyika54). More studies are therefore needed: first to examine the ‘obesogenic’ environment, health beliefs and the social context within which blacks live and work, and second to examine the motivators for behaviour change within this population. Cultural adaptations in interventions involving blacks have also been suggested, such as involving black providers, using a community setting, using the group's preferred language, and incorporating cultural food and activity preferences, traditions and concepts into programme content.
Acknowledgements
This research received no funding or grant from any funding agency in the public, commercial or not-for-profit sector. There is no conflict of interest. G.O.A. did the initial searching in electronic databases and hand searching, and also wrote the first draft of the paper. C.B. did the selection of studies based on the inclusion criteria, and also contributed to the final draft and statistical advice.
