Prevalence and correlates of multiple non-communicable disease risk factors among adults in Zambia: results of the first national STEPS survey in 2017

Introduction the prevalence of Non-Communicable Diseases (NCDs) is increasing in African countries. This study aimed to estimate the prevalence and correlates of multiple NCD risk factors (NCDRF) among the adult population in Zambia Methods nationally representative cross-sectional data from 4,302 individuals aged 18-69 years of the “2017 Zambia STEPS survey” were analysed. Results the prevalence of insufficient fruit and vegetable consumption was 90.4%, followed by overweight/obesity (24.4%), low physical activity (19.5%), hypertension (18.9%), daily tobacco use (10.7%), sedentary behaviour (8.9%), suicidal behaviour (8.5%), alcohol dependence (7.4%), raised total cholesterol (7.4%), and diabetes (6.2%). The distribution of NCDRF was 41.5% 0-1 NCDRF, 48.2% 2-3, 10.4% 4-10, and 26.7% 3-10 NCDRF. In adjusted ordinal logistic regression analysis, compared to persons aged 18-34 years, individuals aged 50-69 years were 3.58 times (AOR: 3.58, 95% CI: 3.95-4.49) more likely to have a higher number of NCDRF. Women were 24% (AOR: 1.24, 95% CI: 1.03-1.49) more likely than men to have a higher number of NCDRF. Persons living in urban locations were 71% (AOR: 1.74, 95% CI: 1.43-2.16) more likely than persons living in rural locations to have a higher number of NCDRF, and compared to individuals with lower than primary education, persons with more than primary education were 20% (AOR: 0.80, 95% CI: 0.65-0.98) less likely to have a higher number of NCDRF. Conclusion more than one in four study participants had three to ten NCDRF and several associated factors were found that can aid to target interventions.


Introduction
In Zambia 29% of mortality was attributed to noncommunicable diseases (NCDs) in 2016 [1]. A high proportion (>80%) of premature deaths from NCDs, such as cardiovascular diseases, diabetes, cancer, and respiratory diseases, occur in low-and middleincome countries [2]. Major behavioural NCD risk factors (NCDRF) increasing the risk of NCD death include tobacco use, unhealthy diet, low physical activity, and hazardous and harmful alcohol use [2]. Considering the increasing trend of NCDs in Africa, it is "crucial to have a careful understanding of the local drivers of NCDs" [3]. Against this backdrop, national data on NCDRF are needed in a lowermiddle income Southern African country, Zambia. Some studies in Zambia were subnational (Lusaka urban district) and only focused on specific NCDRF, such as high cholesterol levels (15.8%) [4], obesity (14.2%) [5], diabetes (4.0%) [6] and hypertension (34.8%) [7]. In two rural districts (Kaoma and Kasama) in Zambia, the prevalence of hypertension was 25.8% and 30.3%, respectively [8]. The adult prevalence of diabetes of 3.5% was found in a household survey in five of ten provinces in 2010 in Zambia [9].

Methods
Nationally representative cross-sectional data from the "2017 Zambia STEPS Survey" were analysed [21]. A "multi-stage cluster sampling technique was used to select a nationally representative sample of adults in Zambia aged 18 to 69 years." [22]. "In the first stage of sampling, Standard Enumeration Areas (SEAs) were selected from each province using a probability proportional to size (PPS), and in the second stage, 15 households in rural SEAs and 20 households in urban SEAs were selected systematically using an appropriate sampling interval based on the number of households in that SEA." [22]. More information on the sampling strategy and the 2017 Zambia STEPS survey data can be publicly accessed; the survey response rate was 74.3%." [22]. "The study was approved by the University of Zambia (UNZA) Research Ethics Committee (REC), and written informed consent was obtained from participants." [22].
Behavioural NCDRF consisted of insufficient fruit and vegetable consumption (<5 servings/day), low physical activity, and sedentary behaviour (=8 hours/day) based on the "Global Physical Activity Questionnaire", daily tobacco use, and alcohol dependence (defined as =4 scores on item 4-6 of the "Alcohol Use Disorder Identification Test=AUDIT" [22]. Suicidal behaviour was based on three questions on suicidal ideation, plans, and/or attempts in the past year [22]. Sociodemographic information included marital status, age, sex, highest educational level, work status, ethnic affiliation, and geolocality [22]. Data analysis: statistical procedures were done with "STATA software version 15.0 (Stata Corporation, College Station, Texas, USA)," taking the multistage sampling design and data weighting into account [22]. The total number of ten NCDRF were grouped into 1=0-1, 2=2-3, and 3=4-10 NCDRF and described with bar graphs and frequency tabulations. Unadjusted and adjusted ordinal logistic regressions were utilized to assess predictors of the number of NCDRF (0-1, 2-3, and 4-10). Co-variates were selected based on a previous literature review [10,13,[18][19][20]. Only complete cases were included in the analysis, and p<0.05 was set as significant.

Results
Characteristics of the sample and NCDRF: the study population included 4,302 individuals aged 18-69 years (31 years median age, IQR 23-41). Almost half of the study participants (48.7%) were men, 59.1% were married or cohabiting, 28.7% were unemployed, 71.1% had primary or more education, 32.8% were Bemba and 48.8% resided in urban areas. The prevalence distribution of individual biological NCDRF was 7.4% raised total cholesterol, 6.2% diabetes, 24.4% overweight/obesity, and 18.9% hypertension. The prevalence distribution of individual behavioural NCDRF was 90.4% insufficient fruit and vegetable consumption, 19.5% low physical activity, 10.7% daily tobacco use, 8.9% sedentary behaviour, 8.5% suicidal behaviour, and 7.4% alcohol dependence. The prevalence of overweight/obesity, low physical activity, raised total cholesterol, and suicidal behaviour was significantly higher in women than in men, and the prevalence of daily tobacco use and alcohol dependence was significantly higher in men than in women (Table 1, Table 2).
The four most prevalent individual NCDRF in this study were insufficient fruit and vegetable consumption (90.4%), overweight/obesity (24.4%), low physical activity (19.5%), and hypertension (18.9%). Similar proportions of individual NCD risk behaviours were found in the STEPS surveys in 2014 in Kenya [10] and in 2013 in Nepal [13]. The prevalence of hypertension (18.9%) in this study was lower than in previous local surveys in Zambia,34.8% in the urban Lusaka district [7], and 25.5%-30.3% in two rural districts in Zambia [8]. The prevalence of overweight/obesity (24.4%) in this study was higher than in the urban Lusaka district study (14.2%) [5], and the prevalence of insufficient fruit/vegetable consumption (90.4%) was higher than in the 2003 Zambia World Health Survey (77.7%) [23], and the prevalence of low physical activity (19.5%) in this study was similar to data from the 2003 Zambia World Health Survey (23.3%) [23].
The prevalence of daily tobacco use (10.7%) and alcohol dependence (7.4%) in this survey was similar to the 2009 Malawi STEPS survey (14.1% smokers, and 7.7% excessive drinkers) [11,15], and the 2014 Kenya STEPS survey (10.1% smokers, and 12.7% harmful alcohol users) [10,14]. The proportions of daily tobacco use and alcohol dependence were similar to results from the 2003 Zambia World Health Survey (14.1% current smoking and 7.4% heavy episodic drinking) [23]. The increase of exercise taxes and prices on tobacco products and alcoholic beverages has been recommended in Zambia [22]. The prevalence of diabetes (6.2%) and raised total cholesterol (7.4%) in this study was similar to the 2009 Malawi STEPS survey (5.6% diabetes and 8.7% raised cholesterol) [11,15], but lower than in the 2014 Kenya STEPS survey in terms of high total cholesterol (10.1%), and higher in terms of diabetes (1.9%) [10,14]. Compared to the prevalence of raised total cholesterol (15.8%) in the Lusaka urban district STEPS survey [4], the prevalence of raised total cholesterol was lower in this study (7.4%), and the prevalence of diabetes (6.2%) was higher in this study than in the Lusaka urban district study (4.0%) [6], and the large community-based study in 2010 in Zambia (3.5%) [9].
Specific NCDRF differed by sex in this study, with substance being higher in men, and overweight/obesity, low physical activity, raised total cholesterol, and suicidal behaviour being higher in women. Similar sex differences in the prevalence of substance use, obesity, and raised total cholesterol were also found in the 2009 Malawi and 2014 Kenya STEPS surveys [10,11]. It is important to take these sex differences into account when designing NCD health promotion activities [11].
Study limitations include the cross-sectional design of the study, which prevents from causative inferences, and the questionnaire interview relying on self-report of the data. Some study variables, such as household income, could not be included in the analysis due to too many missing values.

Conclusion
In this national community-based 2017 STEPS survey among adults in Zambia, more than one in four study participants had three to ten NCDRF. Several factors associated with NCDRF counts were identified, including increasing age, female sex, residing in urban areas, and lower education that can be targeted in interventions to address multiple NCDRF in the Zambian population. Taking the clustering nature of NCDRF into account, interventions should be targeting multiple, in particular modifiable, NCDRF.

Competing interests
The authors declare no competing interests.

Authors' contributions
All authors fulfil the criteria for authorship. SP and KP conceived and designed the research, performed statistical analysis, drafted the manuscript and made critical revision of the manuscript for key intellectual content. All authors read and approved the final version of the manuscript and have agreed to authorship and order of authorship for this manuscript.

Acknowledgments
The data source, the World Health Organization NCD Microdata Repository (URL: https: //extranet.who.int/ncdsmicrodata/index.php/cata log), is hereby acknowledged.  Table 5: multivariable ordinal logistic regression with non-communicable diseases risk factor counts among individuals aged 18-69 years in Zambia, 2017 Figure 1: frequency of non-communicable diseases risk factors among adults in Zambia