Relevance of JD-HDSS
This HDSS in progress is the first in Nepal with focus on health issues. A Household Registration System with HDSS methodology had been in the running in the late 1990’s in the Southern Nepalese town of Chitwan in collaboration with Population Studies Centre [1]. However, that project mainly investigated migration issues and followed even those who had out-migrated from the study area after the start of the project.
The intention of this paper is to share the methodology and initial experience of establishment of a HDSS in Nepal and we aim to present longitudinal data after completion of planned subsequent rounds. In terms of the population size, compared to other HDSSs such as Agincourt in South Africa and Matlab in Bangladesh, JD-HDSS is a modest endeavor in terms of the area covered and population size [6, 20]. Nonetheless, as put forward by Byass and colleagues, there is no exact formula or statistical ways to determine the optimal sample size of a HDSS [5]. Similarly, the external validity of a study is also an important parameter and in that sense, JD-HDSS is a prototype of the villages near the cities in Nepal which are fast growing into peri-urban or sub-urban stature. However, given the ethnic and geographic diversities of Nepal, it cannot claim to be representative of all the ethnic and geological facets of the country. Nevertheless, from the point of view of study objectives, the selection of HDSS area is arguably well suited because historically establishment of a HDSS has always been tailored to the local needs. The HDSS in Western Kenya, for example, was initiated as a field study site for testing the effectiveness of insecticide-treated nets [10]. In that regard, JD-HDSS is optimal in terms of its current focus on urbanization and NCD risk factors, as the chosen villages are rapidly undergoing socio-developmental transitions. And, as shown by our baseline survey from the HDSS, in its early stage of demographic transition, Nepal faces a double burden of diseases. This is characterized by simultaneous infectious disease and an increasing NCD burden due to urbanization and changing lifestyles [21]. Without preventive measures to combat this trend, Nepal will experience a significantly increased NCD burden. Due to current emphasis on curative care rather than preventive measures, Nepal does not identify the root causes of disease (i.e., hidden social, cultural and demographic aspects). Different cross-sectional studies conducted at various intervals have mentioned these risk factors but failed to establish casual relationships between risk factors and disease [22]. Similarly, Nepal has promulgated laws for registration of vital statistics (i.e., birth, death, marriage, divorce and migration) but has not established the necessary reporting mechanisms yet. Consequently, Nepal needs to establish HDSS to monitor different demographic events: birth; deaths; marriage; migration and health events: types of sickness; hospital visits etc. within a defined geographical community, and also provide a podium for interventional studies regarding subpopulations [8, 10].
Nepal has experienced rapid urbanization (i.e., nearly 4.7% per year) [23]. Located near Kathmandu, the capital city, JD-HDSS has already begun the process of shifting from a rural to urban community. Changing lifestyles, from traditional to modern, significantly impact demographic and social structures. Additionally, in- and out-migration impacts JD-HDSS, whose positive net migration rate (0.0085%) indicates population gain. The current baseline survey determined that NCDs are also prevalent alongside communicable diseases, noting that the population, not unexpectedly, has a double burden of disease. Thus, the JD-HDSS study site provides a good platform for studying disease burden from the dual aspects of both health promotion and disease prevention.
Findings from the JD-HDSS baseline study
The crude birth and death rates of the JD-HDSS are 35% and 53% lower, respectively, compared to national figures (CBR =27.7 per 1,000 population/CDR =9.7 per 1,000 population) estimated in the year 2008 [18]. Total age dependency ratio (younger than 15 years and older than 60 years) was 31 per 100 working-age population (15–60 years), of which 24 per 100 belonged to under 15 years of age. Similarly, the baseline survey observed no mortality in infants, children younger than five years of age, and mothers. According to the Nepal Health Demographic Survey 2011, the national infant and under-five mortality rates are 46 and 54 per 1000 live births, but a wide variation has been noted, for example, according to place of residence (urban and rural rates are 38 and 55, 45 and 64 respectively) [24]. Thus the reason for not finding any child or maternal death could be because of improved living standard in the selected study area, easy access to healthcare services (including access to two community hospitals and a health post), the low illiteracy rate, increased income (only 2.4% lower class), and healthier nutrition. It could as well be because of smaller study population.
The study also determined that more than 90% of children were born in health institutions, more than three times the rate disclosed in the preliminary report of the Nepal Health Demographic Survey 2011, where only 58% of mothers received prenatal care from a doctor or nurse and only 28% of all births occurred in health facilities [25]. Although JD-HDSS has met the Millennium Development Goal (MDG) target (i.e., 60% of all births attended by a skilled provider), Nepal itself is far from attaining that goal [16]. About 33% of deaths occurred in individuals younger than 65 years of age in the study area, about two times less than that reported by the 2001 national census [15]. The census also reported higher life expectancy in Bhaktapur district (71.33 years) compared to the Kathmandu and Lalitpur districts (69.5 years and 67.10 years, respectively) and the entire country (64.1 years) [14].
The net migration rate for JD-HDSS was 0.0085% (in-migration = 2.25; out-migration = 1.36). In-migration increased from 13.4% to 26.8% between 1971 and 2001, respectively. Rural-to-urban migration predominates in Nepal, especially in the large cities of Kathmandu, Lalitpur and Bhaktapur districts. According to national census 2001, those districts received 71.8%, 82.7% and 44.6% in-migrants, respectively [26, 27]. The same census showed, the five major reasons for in-migration -business, agriculture, service, education and marriage- mirrored our findings (data not shown) [26, 27] . In our study, the majority (91%) of out-migrated people explained that they migrated to increase earning power and get a better education.
The annual health report for fiscal year 2009–2010 listed gastritis, respiratory problems, and injuries among Nepal’s 10 leading morbidities [18]. About 53.22% of the total population visited the outpatient department in Bhaktapur district, of which 57.27% were male. Among them, 9.22% sought outpatient treatment for skin diseases; 6.17% for gastritis; 3.76% for cardiovascular diseases; 3.18% for dental problems, including pain; 3.70% for headache; 2.57% for hypertension and 1.91% for pain [18]. JD-HDSS observed similar morbidity patterns.
In the JD-HDSS study area, NCD prevalence among individuals older than 30 years was 4.32% associated with several factors, including sex, ethnic group, occupation and education. Currently, few community-based studies have investigated NCD risk factors in Nepal. One study, conducted in Eastern Nepal, reported an NCD prevalence rate of 6% in adult males, and a 2008 survey studying NCD reported that more than 80% of respondents exhibit one or more risk factors [28, 29]. A recent hospital-based study, conducted in 31 health institutions in Nepal, reported that 36.5% of patients suffer from NCD and share of female was 52.4% of the total patients [30]. Our study showed that smoking carries a 26% risk for developing NCD (unadjusted odds ratio1.26 [95% CI: 0.97–1.63]) compared to 60% reported by a recent hospital-based study in Nepal. Similar to our findings, that study also reported that the adjusted odds ratio for smoking was not statistically significant [30].
Ongoing challenges for NCD control in Nepal include political instability, poor health literacy, greater focus on curative aspects and rapid urbanization [31]. The control approach works well if lifestyle factors are modified through effective health education intervention and tobacco control policies. To forecast cumulative risk, future studies should investigate risk factor clusters (i.e., a combination of two or more risk factors) for NCD development rather than focus on single-factor risks [32].
Limitations
Although our success in establishing Nepal’s first HDSS provides a reliable sampling framework for subordinate studies and builds mutual trust between village residents and researchers, the baseline survey includes several limitations that require further attention. Because we obtained our baseline data from any family member who was older than 18 years of age, our data may be both over- and under-reported. There is also a possibility of bias regarding both recall and selection. Selection bias may arise due to the selection of convenient respondents from whom the enumerators can obtain information easily, whereas supervision bias may result from time constraints regarding form checking. Because the JD-HDSS geographic area was selected purposively, our survey covered only two village developmental committees; hence, the findings should be generalized cautiously to populations beyond the surveyed population [33].
To elicit information on death and its cause, the survey interviewer asked the question “Has any member of your family died within the year?” If the answer was yes, respondents were asked about age, sex and the cause of death. So there is possibility of measurement bias, e.g., data enumerators might fail to record all deaths that lead to under reporting of deaths and hence the possibility why the crude death rate in the JD-HDSS is below the national figure (8.3 per 1,000 population). Also, the emotional attachment with the death of a family member may make the respondents not wanting to talk about it. Furthermore, the enumerators do not ask for any details regarding deaths. In addition, there is the possibility of recall bias, i.e., the respondents cannot remember exact date of deaths.
We also found that only 7 of 10 deaths were reported to the village office. It indicates that the vital registration system does not cover all deaths and fails to provide accurate data. This underlines the importance and need of an HDSS in Nepal, which enables recording of deaths on a regular basis. Admittedly, the verbal autopsy method would have been a better tool to elicit the causes of deaths [34]. We were unable to use verbal autopsy for the baseline study because of lack of trained manpower but do intend to use them in future surveys.
Additional interviewer bias can also come from the enumerators as many of them were overqualified for enumeration work and they can add their own thoughts in to a response while recording it. Similarly, the response to the enumerators belonging to a different political ideology can hamper the respondents’ answer during the interview. Additionally, the length of the questionnaire may have contributed hurried and often untrue or incomplete responses. Similarly, lack of ‘on-the-spot’ benefit could have biased the respondent’s reply.
Finally, provision of ancillary care is an ethical dilemma for which there are no clear-cut guidelines or solutions [35]. Over-inflation of the HDSS budget is an important possibility to consider while providing such care [36].
Strengthening of the JD-HDSS
We have used basic technology for village mapping and data collection in the baseline census. Mapping can be done more scientifically in the future with the use of GPS technology like in FilaBavi and many African HDSSs [9, 10]. This could also have paved way for mapping of the environment coverage of the HDSS [37]. Cardiff Teleforms, used for example in Western Kenya, could also be helpful if there is possibility of technical support [10]. Even data entry and analysis software can be upgraded to maintain and produce quality data [6]. Effort should also be made towards improving the internal validity of our data collection technique by using more widely accepted methods such as Verbal Autopsy for information on cause of mortality [6, 38]. Expansion of the study area itself to include a more urban area can yield interesting comparisons in demographic and health parameters [39]. Finally, effort should be made to eventually make this HDSS an IN-DEPTH member.
Future studies
A new round of census survey will begin in September 2012, followed by another survey every six months. JD-HDSS is also currently conducting two longitudinal subordinate studies, one related to tobacco smoking behavior among 500 adolescents and the other related to knowledge, attitudes and practice of cardiovascular disease in people aged 25–59 years. Additional subordinate studies on maternal health and nutritional status are in the pipeline.
In agreement with the thoughts put forward by Baiden, the JD-HDSS can be a good field site for other scholarly interests as the project is primarily of academic nature [40]. Presence of gaps in socio-economic parameters and diversity in health-care utilization, studies on issues such as socio-economic determinants of health, health equity and health system reform are interesting tentative projects in the JD-HDSS [41–43]. As done in the Agincourt HDSS, triangulation of our data with the relevant national data could also be a potentially useful exercise particularly in the context of urbanization and its complexities [44].