The meta-analyses of the prevalence data suggest a rising secular trend of chronic disease in Bangladesh. For both HTN and T2DM, the pooled prevalence estimates were lowest during 1995–2000, highest during 2006–2010, and in-between during 2001–2005 (Figure 4). Compared with the estimates during 1995–2000, those during 2001–2005 were non-significantly higher and those during 2006–2010 were significantly higher. This supports the ‘epidemiological disease shift’ reported previously, which showed an increase in death due to chronic diseases with a concomitant decrease in death due to infectious diseases . The rising trend of HTN and T2DM could be explained, in part, by the increased life expectancy of the Bangladeshi population as well as the increased presence of known risk factors for chronic diseases. Between 1995 and 2010, the mean life expectancy has risen by approximately 6 years (62.1 to 68.3 years) . Also, Bangladeshis rank highest among South Asians in terms of tobacco use, physical inactivity, and poor dietary habits . We anticipated that methodological differences such as sample selection and diagnostic criteria across the studies could influence the estimate of the secular trend. Therefore, we limited our analysis to only those studies that selected a sample from the general population. Further, the cut-off value of systolic (140 mmHg) and diastolic (90 mmHg) blood pressure used to define HTN did not change during the review period. We acknowledge that the diagnostic criteria [43, 44] for T2DM have become less stringent during this time period but this small change alone is unlikely to explain the two-fold rise in the T2DM prevalence.
The HTN prevalence that we found in this review is comparable to the HTN prevalence reported in the other South-Asian Countries, [45–53]. Although a formal comparison among these countries does not fall within the scope of this review, it is worthwhile to compare the data between Bangladesh and India, as they are geographically contiguous, share common traits in culture, and have similar life expectancy and chronic disease risk factor profiles [8, 42]. For example, the HTN prevalence in the rural (pooled: 14.3%) and urban (single study: 22.2%) Bangladesh is well within the range of prevalence estimates for rural (12%–17%) and urban (20%–40%) India . Similarly, the IHD prevalence in the rural areas is comparable between the two countries (Bangladesh = 2.7%–3.4% vs. India = 3%–4%) [12, 20, 21, 48, 54]. The estimates for the urban areas cannot be compared because the Bangladesh estimate (19.6%) comes from only one study, which selected a small sample from a special population (government employees) .
The rising HTN prevalence in Bangladesh is worrisome since a large segment of the population is unaware that they have HTN and since many HTN patients do not have their blood pressure properly managed with medication. Between fifty and eighty-five percent of the participants who were diagnosed with HTN did not know they had HTN prior to the study assessment . Although allopathic anti-HTN medication use (97%) is the norm among those who were diagnosed with HTN, only a quarter of them were found to have their blood pressure controlled with medication. Since uncontrolled HTN is a major risk factor for myocardial infarction (MI), it may partly explain why Bangladeshis have the lowest mean age for the development of MI among the nations of the world .
Similar to HTN and IHD, the T2DM prevalence in Bangladesh is also comparable to the corresponding estimates from other South-Asian countries such as India . Although, in both countries, the prevalence is low in the rural areas (Bangladesh = 5.1%, India = 3.8%), it is considerably higher in the urban areas (Bangladesh = 10.2%, India = 11.8%) . At the same time a significant portion of the population have conditions such as IFG or IGT, and therefore are at risk of developing T2DM [14, 30, 35]. For example, in our reviewed studies, the pooled prevalence for IFG and IGT were found 7.0% (95%CI: 3.2%–10.8%, I2 = 99%) and 8.2% (95%CI: 3.9%–12.4%, I2 = 99%), respectively [10, 16, 22, 25, 30, 31, 33, 35, 36].
Of all the factors that may be contributing to the rise of HTN and T2DM in Bangladesh, behavioral risk factors are likely to be the most salient. A recent review suggests that 90% of the chronic diseases among the South-Asians are attributable to life-style factors . Survey-data indicates that between 47% and 63% of men and less than 3% of women in Bangladesh are daily smokers [14, 55–57]. A significant portion of these smokers uses filter-less cigarettes. Further, smoke-less tobacco use (with bettle-leaf or chewed/sniffed in the powdered form) is quite common (30%) in the population, particularly among the poor and in rural areas, irrespective of gender . Bangladeshis also rank highest among the South Asians in terms of physical inactivity and low intake of fruits and vegetables [58, 59]. And, although the overweight prevalence is relatively low (men = 10%, women = 15%), a large portion of Bangladeshis are, in fact, centrally obese (men = 25%, women = 60%) [34, 35, 60].
Strengths and limitations of the reviewed studies
It is notable that dozens of studies have been conducted in the last fifteen years to assess the risk factors and overall burden of chronic disease in Bangladesh, despite it being one of the poorest countries in the world and having little allocation of government funding for research. Many adopted a unique sampling strategy, in which they selected a village or a group of villages and enrolled all eligible participants. This particular strategy ensured that each study, besides having a robust sample, also had adequate representation in age, gender, and socioeconomic strata. Studies were also conducted in special populations such as pregnant women , health care professionals , government employees , and slum dwellers . Inclusion of both urban and rural samples enabled a few studies to make a direct comparison of disease prevalence and risk factors between these two groups [14, 55]. Objective measurement (such as blood pressure measurement, echocardiogram, or blood analysis) of disease outcomes was common to most studies. Further, studies clearly defined the eligibility criteria and included relevant covariate data.
A general limitation of the reviewed studies was a lack of representative study samples in the urban areas. For example, the studies that were conducted in the capital have recruited participants either sharing a common living environment  (e.g. slum dwellers) or a common profession  (e.g. health care workers). Such samples are unlikely to provide a true estimate of CVD or T2DM prevalence for a city with 12–15 million residents living in a wide range of socioeconomic and environmental conditions. On a similar note, using a single or a group of villages in a certain geographic location may not capture the disease prevalence for a country with approximately 68 thousand villages.
Strengths and limitations of the review
By categorizing the studies according to the markers of good study design (random sample, sufficient sample size, objective measure), we showed that the disease prevalence varied by the study quality. The prevalence was generally higher in ‘high-quality’ studies than in ‘poor-quality studies’, for both HTN and T2DM. We conducted meta-analyses to calculate the pooled prevalence of disease. However, data interpretation should be done cautiously since the heterogeneity (between-study variability) was very high, despite the use of random effects models to accommodate this variability. A limitation is the small number of studies, which have prevented us from stratifying the time trend analysis by study quality.
A general difficulty of this project was to gather all the published data on CVD and T2DM for the review period, since not all Bangladeshi journals are available electronically, and those that have electronic text came into existence recently. Although we were able to collect a number of full text articles that were otherwise unavailable in the Web through personal communications with Bangladeshi researchers, we had to rely on the abstract for a few studies. In most cases, the information needed for the quality rating was available in the abstract and in the two instances where it was missing, we have imputed with the lower (i.e. more conservative) value. There may also be unpublished disease prevalence data that we have missed in our review.
Although an adequate number of HTN prevalence studies have been conducted, they were mostly done in and around Dhaka city, leaving many regions of the country unstudied. Little data exist on the prevalence of other cardiovascular diseases such as ischemic heart disease. Further, we did not find prevalence data on rheumatic heart disease or tuberculous pericardial disease, despite the fact that many Bangladeshis, especially the slum dwellers, suffer from poor nutrition and live in unsanitary conditions -- both of which have been associated with the development of these diseases. No study has reported to date the impact of nutritional (i.e. sodium, refined sugar intake), occupational, and various environmental factors (i.e. migration to the city, salination of drinking water, air pollution) on CVD or T2DM. It is important that these issues be explored for a number of reasons: (1) the urban population has been growing on an average of 3.5% a year , (2) food and drinking water in the coastal area has been greatly affected by a rise in salinity [62, 63], and (3) the urban areas have seen a decline in air quality .
We recommend that future prevalence studies focus on the gaps identified in this review. In addition, we suggest that samples of the future studies are selected systematically, according to the administrative structure of the country (division, district, sub-district, etc.), in order to capture any differences in disease prevalence by geographical regions. We did not find any longitudinal cohort studies on CVD or T2DM. This is a significant gap in the knowledge and understanding of these chronic diseases in the context of Bangladesh. Such studies would provide essential information on the incidence of these diseases, their associated risk factors, and the groups that are at higher risk of developing them. Further, longitudinal data are necessary to understand disease progression and prognosis.
We recognize the potential challenges to conducting cohort studies in Bangladesh. Long-term follow-up of participants may be difficult. In the urban areas, many people are either renters or slum dwellers who change their residence frequently. A similar challenge may arise in rural areas as there is constant migration to the cities for employment. Other challenges include the absence of either disease/death registries or electronic medical records. We acknowledge that the development of a cohort study would require considerable effort and data would not be available for many years. In the meantime, retrospective cohort studies could be conducted using medical records available in various government (NICVD-National Institute of Cardiovascular Diseases) and private institutions (ICDDR,B-International Center on Diarrheal Diseases and Research, Bangladesh; BIRDEM-Bangladesh Institute of Research and Rehabilitation for Diabetes, Endocrine, and Metabolic Disorders; NHF-National Heart Foundation). Such retrospective studies could begin to address some of the research issues outlined above such as disease complications and treatment adherence.