THE PREVALENCE OF CHRONIC CONDITIONS ASSOCIATED WITH MODIFIABLE HEALTH RISK FACTORS IN CORPORATE EMPLOYEES IN SOUTH AFRICA

The Sunday Times Discovery Healthiest Company Index Survey collected a dataset of 13 578 responses from corporate employees in 101 companies. This dataset has been used to assess the prevalence of self-reported modifiable health-risk factors and the association with chronic conditions. The analysis indicates that there is a greater likelihood of an individual reporting a chronic condition where their lifestyle risk factors are outside of the healthy range. This suggests that there are opportunities to manage the costs of health insurance, medical expenses and productivity losses by applying risk management tools aimed at addressing modifiable risk factors.


1.1
The World Health Organization (WHO) has noted that the majority of deaths and disability cases worldwide are a result of chronic conditions. 'Chronic' refers to a condition that persists for longer than three months or is recurrent. 1 1.2 Chronic diseases, particularly non-communicable chronic diseases (NCDs) pose a large and growing global health problem. These diseases include cardiovascular disease, type-2 diabetes, certain cancers, chronic lung disease and depression. They are the leading causes of morbidity and mortality, not only in developed countries but even more so in developing countries (Mayosi et al., 2009).

1.3
The burden of NCDs in sub-Saharan Africa reveals that the epidemiological transition from infectious diseases such as HIV/AIDS and tuberculosis to NCDs is well under way (Dalal et al., 2011). A quarter of all deaths in the sub-Saharan Africa region were due to NCDs in 2004, and this is projected to increase to 46% by 2050 (Dalal et al., op. cit.). Furthermore, the burden of NCDs based on age-standardised death rates is currently higher in the Democratic Republic of Congo, Ethiopia, and South Africa than that in high income countries (Dalal et al., op. cit.).

1.4
Given the burden of NCDs both internationally and in South Africa, there is increasing attention being paid to prevention efforts focused on addressing modifiable risk factors. The World Health Assembly's Resolution on Diet, Physical Activity and Health highlights the need to reduce the risk factors for NCDs that arise primarily from inactivity and unhealthy diets. 2 This resolution encourages member states to increase awareness of the impact of these risk factors and the potential benefits of preventative intervention programmes. Similarly, the South African Declaration on the Prevention and Control of Non-Communicable Disease 3 promulgated and signed in September 2011, reflects the need to address lifestyle factors that contribute to NCDs. Thus there is a global, national and regional commitment to reduce the burden of NCDs.

1.5
In addition to the human costs of NCDs, the economic burden is well documented (Lim et al., 2013, Murray & Lopez, 2013 Indeed, in the World Economic Forum's Executive Opinion Survey more than half of business leaders across low-, middle-and high-income countries expressed concern that NCDs will adversely affect their bottom line in the next five years (Mayer-Foulkes, 2011).
1 Global Burden of Disease Estimates. World Health Organization, Geneva, 2002a 2 Regional strategy and plan of action on an integrated approach to the prevention and control of chronic diseases, including diet, physical activity and health. 138th Session of the Executive Committee of the World Health Organization, Provisional agenda item 4.9 CE 138/17, 2006a 3 Participants in the South African Summit on the Prevention and Control of Non-Communicable Diseases in Gauteng from 12-13 September 2011 1.6 From an actuarial perspective, escalations in the prevalence of chronic conditions places upward pressure on the cost of insuring the mortality and morbidity risks associated with such conditions. Actuaries are involved in mortality and morbidity forecasting as well as calculations to test the financial benefits of treatment protocols associated with chronic conditions under health insurance products. Understanding the relationship between modifiable health risk factors and chronic conditions will assist actuaries with pricing and risk management.

1.7
There is relatively little research in this area in South Africa, especially in relation to the South African employed population. This represents a gap in the literature as the South African employed and insured population is of particular interest in the context of disease management. As interventions need to be evidence-based, this paper aims to investigate chronic disease prevalence and modifiable lifestyle factors in a sample from this population, as well as to assess the association between modifiable lifestyle risk factors and chronic conditions in the studied population. This evidence may be helpful in targeting the implementation of disease management programmes.

1.8
Section 2 of the paper discusses the risk factors that are associated with NCDs and their effects. Risk management opportunities and programmes are then addressed in Section 3. Section 4 describes the Discovery Health Healthy Company Index (DHCI) survey and the results are set out in Section 5. Section 6 provides a comparison of results to other sources leading to a discussion and conclusion in Section 7.

2.1
Healthy lifestyle behaviours, which include regular physical activity and good nutrition, are associated with longevity. 4 Risk factors such as tobacco use, lack of physical activity and unhealthy diet are associated with cardiovascular disease, diabetes and cancer (Mayosi et al., op. cit.). As poor diet, insufficient physical activity, alcohol and smoking have been identified as risk factors for non-communicable diseases, the prevalence of these diseases can be attenuated by modifying these lifestyle behaviours (Steyn, 2006). Although the significance of each risk factor varies by region and culture, Khatib (2004) suggests that on a global scale, the modifiable factors account for 75% of the disease burden. Patel et al. (2010) identified the presence of a chronic condition of lifestyle as being a primary factor for predicting hospital admission rates.

2.2
2.3 NUTRITION 2.3.1 Fruit and vegetables are an essential part of a balanced diet. 5 They help to prevent many diseases and cancers-particularly those relating to the digestive tract-because they contain antioxidants, fibre and nutrients which suppress the action of carcinogens. 6 Results from analyses conducted by the WHO suggest that almost 2,7 million deaths annually can be attributed purely to low fruit and vegetable intake. In developed countries, this accounts for 3,9% of the total disease burden. 7 2.3.2 Extensive advertising campaigns and marketing techniques have made fatty, refined, sweet and salty foods more profitable than healthier alternatives, and this finding has led to fast food outlets becoming increasingly popular in developing countries (Steyn, op. cit.). In South Africa, and other African regions, the African diet has been sidelined by the western diet, resulting in a change in the nutrition patterns of Africans. This change has been documented in studies by Tanner & Lukmanji (1987) and MacIntyre et al. (2002). They found that the traditional African rural diet comprises of a high carbohydrate component, low fat intake and large amounts of plant protein. This is in contrast to western diets which are associated with high levels of saturated fats, high levels of sugar, salt and alcohol intake and low fruit, vegetable, fibre and vitamin intake.
In developed countries, it is estimated that 9,2% of the disease burden arises from alcohol use. According to the 2003 South Africa Demographic Health Survey (SADHS), 21,4% of men and 6,9% of woman aged between 15 and 64 had problems related to alcohol consumption (Mayosi et al., op. cit.).

2.4.2
According to the South African Department of Health 'sensible drinking' is defined as three or two standard drinks per day for men and women respectively, where a standard drink is defined as a can of beer, one tot of spirits, one glass of wine or one small glass of sherry (Vorster, 2001). Serfontein et al. (2009) conducted a sample study of the African population in the North West province of South Africa. They found that found that 18% of their sample population consumed more than 30g of alcohol per day for males and 6% of the study population consumed more that 15g per day for females and hence were defined as heavy drinkers, but there is little research on the prevalence of alcohol consumption within the urban employed population in South Africa.
2.5.2 Between 20% and 25% of all South African adults are smokers (Ayo- Yusuf, Swart & Pickworth, 2004;Steyn et al., 2002). Campaigns and anti-smoking legislation appear to have had some effect. In surveys implemented by the South African Advertising and Research Foundation, it was found that smoking rates in adults fell by twenty percent in the period from 1995 to 2004 (Saloojee, op. cit.).

2.6
PHYSICAL INACTIVITY 2.6.1 An estimated 6% of the burden of coronary artery disease and 7% of type-2 diabetes is due to physical inactivity (Yusuf et al., 2001). Lee et al. (2012) has estimated that 9% of premature mortality is due to physical inactivity and if the prevalence of inactivity decreased by 10%, more than 533 000 deaths could be avoided per year. Members of the WHO's ministerial round table sum up the situation eloquently: Advances in telecommunications and transportation have minimised the need for physical activity. Ironically, while technology has reduced the time it takes to perform specific tasks, use of these mechanisms has been accompanied by an increasing perception of lack of time or lack of control over time for activities that are beneficial to health. Hence, it is easier to ride a vehicle than to walk. 9 2.6.2 A review of recent research in this area shows that there is a dose response between the amount of physical activity and the risk of coronary artery disease (Sattelmair et al., 2011). Individuals who met the physical activity guidelines of 150 minutes per week had a 14% lower risk of coronary artery disease, while those who exercised more than 150 minutes per week had a 20% reduced risk (Sattelmair et al., op. cit.).
2.6.3 In South Africa, 44,7% of men and 47,6% of women are insufficiently physically active and not meeting the guidelines for health-enhancing physical activity (Guthold et al., 2008)

3.1
Insurers have attempted to contain costs related to chronic conditions by benefit designs and by offering differential plan types. These measures include the application of evidence-based algorithms for approving chronic medication, the use of generic and formulary drugs, co-payments for chronic medication and deductibles for certain procedures (Jurisich & da Silva, 1996). Combined with these measures, insurers and health -maintenance organisations have sought to manage the costs associated with chronic and high-risk patients by implementing disease and case management programmes ( Chapman, Lesch & Baun, 2007).

3.2
Disease-management programmes seek to improve compliance and selfmanagement and thereby improve outcomes and long-term costs. While many of these programmes have shown improvement in clinical care and health outcomes there is no definitive proof that they are effective. One reason for this may be that programmes managed by disease-management companies tend to be separate from primary care and focus on providing care for high-risk patients (Bodenheimer, 1999). These programmes thus tend to be less concerned with prevention initiatives and lower-risk patients (ibid.).

3.3
Some insurers have recently become conscious of the need to shift from simply containing costs and managing disease to improving the primary health of members. This shift encourages plan members to adopt healthier practices offered by preventive and wellness programmes (Patel et al., 2010). This is based on research evidence relating to cost saving. For example, Pronk et al. (1999) found that members of a health insurance plan who never smoked, with a mean BMI of 25kg/m² and who participated in physical activity three days per week had mean annual healthcare charges that were approximately 49% lower than physically inactive smokers with a BMI of 27,5kg/m². They argue that funders seeking to minimise health care charges may wish to consider strategic investments in interventions that modify adverse health risks. Similarly, Wolf & Colditz (1998) compared healthcare expenditure, over a one-year period, between a lifestyle intervention group and a usual care group in obese type-2 diabetic patients who were members of a health insurance plan. They reported that after taking into account the costs of a lifestyle intervention programme the total costs were US$3 586 per person per year less among the intervention group compared to the usual-care group for an intervention cost of US$328.

3.4
Despite significant expenditure on managed care initiatives, South African medical schemes have reported an escalation in the prevalence of chronic conditions amongst covered beneficiaries. The Council for Medical Schemes (CMS) reports that the top 10 chronic conditions have shown the fastest increase in prevalence from 2006 to 2011. 10 This escalation may be influenced by the requirement to cover PMBs at cost. The top 10 conditions are given in Table 1.

3.5
Thus it is clear that insurers¸ and the actuaries that advise them, have a direct interest in managing the financial burden of chronic illness in the workplace by addressing modifiable risk factors. In order to do this effectively however, it is necessary to obtain more information on chronic conditions in the workplace, as well on the prevalence of risk factors in the specific population of interest. In the following sections, the authors report on the DHCI survey which provides information on the prevalence of chronic conditions, risk factors and the association between them, in a large sample of South African corporate employees. The authors also compare these findings, which are based on self-reported data, with reference data from the risk equalisation fund (REF) data in order to establish some level of validity in relation to the self-reported data.

4.1
The DHCI was a national initiative seeking to promote worksite wellness and to identify the healthiest companies in South Africa. The campaign was initiated by Discovery Health, the administrator of the largest open medical scheme in South Africa, in partnership with the Sunday Times, a South African national newspaper (Patel et al., 2013).

4.2
The methodology for conducting the DHCI received clearance from the humanresearch ethics committee (medical) of the University of the Witwatersrand in 2010.

4.3
A media campaign informing South Africans of the DHCI and inviting worksites to register was launched in October 2010. The human-resource managers of existing corporate clients of Discovery Health were invited via email to participate in the DHCI. A website-www.healthycompanyindex.co.za-was established to publicise the campaign and to register companies that wished to participate. Participation was not limited to existing Discovery Health clients.

4.4
Human-resource managers or similar company representatives then enrolled their company in the Health Company Index survey website. Once registered, they received an email from the research team, confirming participation. The human-resource manager was requested to send their employees an email informing them of the survey and inviting them to participate in the DHCI. Confidentiality and the voluntary nature of participation were emphasised. Employees were reassured that their employers and Discovery Health would not have any access to their data, as all results were encoded and collated with no personal identifiers.

4.5
Following the information from the human-resource manager, employees received an email inviting them to register their participation on the DHCI website, and to complete the employee questionnaire within two weeks of the invitation. This questionnaire was accessed using a protected online domain. The employee invitations were sent out in bi-weekly batches from 31 January 2011 to 28 February 2011 and the survey was closed on 15 March 2011. The questionnaire is included in the Appendix for completeness.

4.6
In order to ensure a minimum level of statistical credibility only companies with 50 or more employees with internet access were eligible to participate in the DHCI. Participating employers where the response rate exceeded a certain threshold received a report benchmarking the health risk factors of their employees who participated in the survey with the full set of survey respondents. For employers with between 50 and 80 invitations sent, the threshold response rate was 30% of invitations sent while for participating employers with 80 or more invitations sent, a minimum of 24 responses was required. Those employers where the number of responses gave a maximum Bernoulli error of 5% were eligible for rating in the DHCI. The formula for calculating the Bernoulli error is given in equation 1: where n represents the number of email invitations sent to employees of that employer.

4.7
The manager responsible for the company's workplace health management programme or the human-resource manager completed an online employer questionnaire, which included sections on current health-promotion initiatives, on-site facilities, company health-related policy and leadership support. The survey was largely based on the National Worksite Health Promotion Survey (Linnan et al., 2008) which had similar aims.

4.8
The employee questionnaire was primarily a health-risk assessment in which employees self-reported clinical measures including blood pressure, height, weight and cholesterol results. The employer questionnaire was used to assess only which interventions were available in the workplace, and was not used in any further analysis. For the purposes of the analysis presented in this paper, the employee questionnaire was used to assess what chronic conditions were present. This was assessed by asking: Has a doctor told you that you have any of the following conditions? If yes, please indicate whether or not you are currently on medication.
The conditions listed were heart disease, high blood pressure, diabetes, high cholesterol, cancer, chronic lung disease (for example, emphysema, chronic bronchitis), and depression. The presence of the conditions was therefore self-reported. There was also no indication provided of the severity of the condition or the type of treatment used for chronic conditions. The questionnaire did not distinguish diabetes type. The employee questionnaire did not include questions on medical scheme membership.

4.9
Self-reported lifestyle behaviours such as smoking status, habitual levels of physical activity and nutrition habits, and the employee's willingness to improve these measures were included in the employee questionnaire.
4.10 101 companies participated in the survey and email invitations were sent to 58 737 employees. There were 13 578 employees who completed the employee questionnaire. This represented a response rate of 23,1%. The data presented in this study were collected solely from the employee questionnaire.
4.11 Since the data from this survey was sourced from email invitations, the scope is limited to employees with internet access and this means that the results may not be applicable to the broader South African workforce.

5.1
The average age of the responding employees was 36,4 years at the date of survey and 59,2% were female. Age and gender were used as the key risk factors for segmenting the population. The distribution by age and gender for all respondents is shown in Table 2.

5.2
Of the employee respondents, 3 299 reported having at least one of the chronic conditions listed in the question given in ¶ 4.8 and were termed 'chronic' for the purposes of this paper. Of the employee respondents, 765 reported having two or more of the conditions and the distribution of these responses is shown in Tables 3 and 4.

5.3
Taking into account multiple conditions, there were 4 281 cases reported across the 13 578 respondents and the distribution thereof is shown in Table 5. The employee questionnaire included questions relating to various risk factors that have been shown to affect an individual's health status. These included: -Body mass index (BMI) defined as weight in kilograms divided by the square of the height in metres; -Physical activity measured as minutes of exercise per week at moderate intensity; -Smoking status, defined as current smoker, past smoker or never smoked; -Nutrition measured in terms of fruit and vegetable servings per day; and -Alcohol measured as units of consumption per week.

5.4.2
The responses per risk factor were classified as low, medium or high risk as set out in Table 6. The low risk category is considered as being 'within the healthy range' and the medium and high risk categories are considered to be 'outside the healthy range' (Kolbe-Alexander et al., 2008).

5.4.3
The proportion of respondents who fell outside of the healthy range for alcohol consumption as defined in ¶5.4.2 was very low (2,64%) and so no further analysis was done on this factor. This may be attributable to the self-reporting nature of the questionnaire or binge-drinking behaviour, which was left for future analysis by the authors.

5.4.4
The comparison of the proportion of respondents who fell within the healthy range for each of the risk factors is shown in Table 7. The comparison of the proportions in the chronic group and lives with no self-reported chronic conditions is made using a two-sample sign test (Mood, Graybill & Boes, 1974). All of the risk factors defined in ¶5.4.1 except nutrition, were significantly different ( p<0,001) for the chronic group compared to the non-chronic group. The non-significance of the nutrition factor may be due to the simplified measurement in terms of fruit and vegetable intake only. Further work has been done on defining this factor subsequent to the first DHCI survey. Significance *** *** *** *** p < 0,001, ** p < 0,01, * p < 0,05 5.4.5 The relative distribution by age group and risk category is shown graphically in figures 1 to 4. It is interesting to observe that the patterns by age are different for the various risk factors and that for smoking and nutrition, the prevalence of high-risk behaviour tends to reduce with age. 5.4.6 The relative proportions of respondents classified as being outside the healthy range for each risk factor were also considered for each condition. The results are shown in Table 8. The levels of significance are affected by the smaller sample size when the respondents are divided into different chronic conditions.    The respondents reporting a heart condition had greater proportions of respondents outside the healthy range for smoking ( p<0,001), BMI ( p<0.01) and physical activity ( p<0.05) than respondents not reporting a heart condition.

5.4.8
Similarly the respondents reporting high blood pressure-the most prevalent condition-had greater proportions of respondents outside the healthy range for BMI ( p<0,001) and physical activity ( p<0,001) than respondents not reporting high blood pressure. 5.4.9 The respondents reporting diabetes had greater proportions ( p<0,001) of respondents outside the healthy range for BMI and physical activity. For smoking the proportion outside the healthy range was greater than for non-diabetics but with a lower level of significance ( p<0,05). It is interesting that the nutrition factor is not significant as dietary changes are associated with diabetes treatment.
5.4.10 BMI ( p<0,001) and smoking ( p<0,05) were also associated more with respondents reporting high cholesterol than respondents without high cholesterol. The proportion of respondents reporting high cholesterol but in the healthy range for nutrition was greater ( p<0,05) than for respondents not reporting high blood pressure. This may indicate some behavioural change by these respondents due to their condition.

5.4.11
The respondents reporting cancer only had a greater proportion in the medium-and high-risk categories than the balance of respondents for smoking ( p<0,05). 5.4.12 Behaviour outside the healthy range for smoking ( p<0,05), nutrition ( p<0,05) and exercise ( p<0,05) was also more prevalent among respondents reporting chronic lung disease than respondents not reporting this condition. Chronic lung disease was the condition with the lowest prevalence and so these results are based on a very small sample and should be interpreted with caution.

5.4.13
The respondents reporting depression were more likely to be outside the healthy range for physical activity ( p<0,001) and smoking ( p<0,001) than respondents not reporting depression.
5.4.14 The association between the individual risk factors noted above and the chronic conditions is assessed using odds ratios in Table 9. These indicate that, for example, having a BMI outside of the healthy range is associated with a 1,94 greater likelihood of reporting at least one chronic condition. It was not possible to calculate the odds ratios for co-morbidities due to the small sample size.

5.4.15
The odds ratios per chronic condition and risk factor are shown graphically in Figure 5. The association between BMI being outside of the healthy range and increased likelihood of high blood pressure and diabetes is evident as well as the association between lower physical activity and increased likelihood of diabetes and chronic lung disease. Only for nutrition are there any odds ratios less than 1.

5.4.16
In order to examine the association between the risk factors and the chronic conditions a logistic regression was performed, including all risk factors and covariates to normalise for age and gender. The standard Wald hypothesis test (Hosmer & Lemeshow, 2004) was used to assess the likelihood of each risk factor being outside the healthy range for each condition. The significant factors are shown in Table 10. For  Figure 5. Odds ratios per chronic condition and risk factor example, respondents with cancer were 1,66 times more likely to be smokers than respondents not reporting cancer and respondents with diabetes were 2,872 times more likely to have a BMI outside of the healthy range than non-diabetic respondents. The table only shows the risk factors that were significant and BMI and smoking feature the most across the diseases followed by exercise and nutrition.

5.5
CHRONIC CONDITIONS AND SELF-REPORTED HEALTH STATUS 5.5.1 Respondents were asked to classify their state of health and the responses are shown in Figure 6. Of the respondents reporting at least one chronic condition, 15,4% indicated that they consider their health to be excellent as compared to 30,6% of the non-chronic group which is significant at the 0,01% level. Of the respondents in the chronic group, 3,6% classified their health as poor or very poor as compared to 1,1% in the non-chronic group which is significant at the 0,01% level. 5.5.2 It can be challenging to convince an individual that their lifestyle choices are affecting their health status and that these can be modified (Lim et al., op. cit.). The results of the health perception question indicate that respondents in the chronic group are more aware of the adverse implications on their health than the non-chronic group. However 40,7% of chronic respondents consider their health to be good or excellent and are therefore less likely to adjust their behaviour according to Lim et al. (op. cit.).

5.6
CHRONIC CONDITIONS AND SCREENING TESTS 5.6.1 Respondents were also asked about whether they had undergone screenings for blood pressure, cholesterol, glucose and HIV in the previous 12 months and to indicate the results of these screenings. HIV was included due to the importance Figure 6. Perception of health by number of self-reported chronic conditions of HIV risk awareness in the South African population. The respondents reporting at least one chronic condition were more likely to have undertaken all the screening tests ( p<0,001) than the non-chronic group except HIV where they were less likely to have had a test ( p<0,001). These results are shown in Table 11.

5.6.2
The respondents who had taken a test in the last 12 months and who could remember the result were asked to indicate the result. The classifications of the test results into low, borderline and high for the blood pressure, cholesterol and glucose tests were defined as per Table 12. Figure 7 shows that there was a higher prevalence of high and borderline results among chronic respondents than among non-chronic respondents. This effect was less marked for glucose.  Cholesterol Below 5,0 mmol/l 5,0 mmol/l to 6,0 mmol/l 6,1 mmol/l or higher Glucose Below 7,8mmol/l 7,9 mmol/l to 9,9 mmol/l 10 mmol/l or higher 6. COMPARISON OF RESULTS TO OTHER SOURCES 6.1 South African medical schemes report on the prevalence of 26 chronic conditions among their beneficiaries to the Registrar of Medical Schemes as part of a shadow process for implementing a REF. The clinical criteria for classifying a beneficiary as having a chronic condition are defined clinically by the CMS. 11 These criteria therefore require clinical diagnosis as compared to the self-reported nature of the DHCI.

6.2
A prevalence of medical scheme beneficiaries per chronic condition was published by the CMS in 2007 based on a study conducted across the four largest administrators. 12 The treated case prevalence per 1 000 beneficiaries for the conditions comparable to those addressed in the DHCI survey are shown in Table 13. Since type-1 and type-2 diabetes were not distinguished in the DHCI questionnaire, the DHCI diabetes statistics have been compared to the REF type-1 and type-2 diabetes combined statistics. revealed that none were statistically significantly similar. This adjustment takes account of age-based differences in the sample population but not the self-selection effect and the fact that the sample includes only employed people with internet access who may have different predisposition that the general population. The comparisons per age group are shown in Figure 8 and this is extended to condition level in Figure 9. 6.4 The higher prevalence of all chronic conditions except for chronic lung disease in the DHCI group relative to the non-chronic group may reflect the effect of self-reported data as well as the fact that REF eligibility criteria refer to a level of treatment rather than just a diagnosis. The self-reported treatment data in the DHCI data was not specific enough to allow a comparison to the REF eligibility criteria.

6.5
The South African Health Review (Padarath & English, 2013) provides national statistics on hypertension as a proxy for chronic disease indicators. These statistics are

6.6
The Actuarial Society of South Africa produced a discussion document in 2010 13 that used the SADHS conducted by the Department of Health to compare population prevalence to medical scheme prevalence for certain conditions including hypertension, asthma, type-1 and type-2 diabetes and hyperlipidaemia. The SADHS prevalence is also self-reported and was compared to the REF prevalence which is treatment based. Hypertension and asthma were found to have a higher prevalence in the general population than in the medical scheme population; diabetes had similar prevalence and hyperlipidaemia had much lower prevalence in the general population. While this paper reported on an association between the prevalence of selfreported chronic conditions among respondents and their health risk factors, it is a crosssectional analysis and so no causal relationships can be confirmed. A further limitation of this study is the reliance on self-reported data from a population of employees with internet access. The results therefore relate to the employed population who are likely to be in administrative, sales, management and professional roles rather than a representative spectrum of the employees. The prevalence of chronic conditions for this group may be significantly different to that of the general population.

7.2
The results have demonstrated that the prevalence of certain modifiable risk factors in an employed population can be used as an indicator of the prevalence of NCDs and the propensity for these conditions to develop.

7.3
The proportion of individuals with modifiable health risk factors outside of the healthy range is higher amongst those DHCI respondents that reported having at least one chronic condition as compared to the balance of respondents, except for nutrition. This suggests that there is an opportunity to encourage healthier lifestyle practices among employees in order to contain the risks associated with chronic conditions.

7.4
The respondents who reported high cholesterol had a higher proportion ( p<0,05) in the healthy range for fruit and vegetable consumption which suggests that they may have already changed behaviour as a result of their chronic condition.

7.5
The modifiable risk factors of smoking and BMI had the highest significance for respondents in the chronic group. Higher risk BMI is often a consequence of poor nutrition and a lack of physical activity and so it is important to note that these factors are not necessarily independent.

7.6
It is not clear whether individuals reporting chronic conditions are aware of the potential positive effects that modifying their behaviour can have on their health. There appears to be a particular need to promote behavioural change in smoking and levels of physical activity, however it appears that the challenge is to convince people, particularly those who do not have a chronic condition but are at risk of developing one, that their current lifestyle choices are affecting their long-term health.

7.7
The DHCI survey data indicates a higher prevalence of chronic conditions associated with lifestyle conditions than the REF data. This may be attributable to selfreporting but indicates that there is a risk of such conditions developing to the point that they meet the clinical criteria required by REF.

7.8
NCDs are not only associated with increased medical costs. The prevalence of chronic conditions of lifestyle in an employee population has implications in terms of organisational risks, such as absenteeism, treatment costs and productivity (Goetzel et al., 2003). The prevalence of chronic conditions of lifestyle amongst DHCI respondents suggests that there would be associated organisational risks in the South African employed population. These would have cost implications for the employers and for insurers covering this population.

7.9
Measuring the levels of behaviour with respect to modifiable health risk factors using questionnaires similar to that employed for the HCI provides an opportunity to identify the propensity for developing NCDs in a population. While this is useful for actuarial risk assessment, there is also an opportunity to intervene with programmes targeted at higher risk individuals rather than general health education.
7.10 Many disease management programmes currently adopted by employers and medical schemes are based on treatment of existing chronic conditions by providing access to medication and monitoring rather than promoting behavioural change. The literature noted above and the findings suggest that there is an opportunity to prevent the development of NCDs by encouraging employees to modify their risk behaviours. 7.11 This is a multi-stage process of creating awareness of this risk associated with certain practices and then providing opportunities for behavioural change (Beaglehole et al., 2012). Many of the challenges with reducing NCD prevalence identified by Beaglehole et al. (op. cit.) relate to behavioural change and a survey such as the DHCI contributes to raising public awareness , modifying risk factors and engaging businesses and the community, all of which are goals identified by (Beaglehole et al., op. cit.). Patel et al. (2013) have also identified the workplace as a site where there is an opportunity to promote behavioural change with respect to modifiable risk factors rather than only leaving this to the realm of the medical schemes.
7.12 Based on the employer survey, 45% of the companies that participated in the DHCI reported having a comprehensive onsite wellness programme and 39% of the participating employers reported having a disease management programme. While employees in the balance of participating companies may have access to disease management programmes through their medical scheme, the employer may be able to encourage greater enrolment in such programmes by offering them directly, possibly in co-operation with their medical schemes. An area for further work is to assess whether the employees reporting chronic conditions who are employed at a worksite where a disease management programme is offered, have lower risk factors and better health outcomes than the balance. 7.13 The DHCI data indicate that the self-reported prevalence of chronic conditions among employees in participating companies is associated with modifiable health risk factors. This suggests that there is an opportunity for South African employers to be more active in educating employees on the benefits of modifying their health risk behaviours and the literature indicates that there is a potential economic benefit from investing in such disease management programmes.
Which province do you work in? Are you able to work flexi-time, that is are you able to work at times convenient to you (for example, before or after the normal workday)?

Yes No
Are you mostly: Office-based Remote/field-based Both Is your position: In the past 12 months how many days of annual holiday leave have you taken? 0 days 1-2 days 3-4 days 5-6 days 7-8 days 9-10 days 11-13 days 14 days + In the past 3 months how many times did you visit the doctor? 0 1 2 3 4 5+

Your alcohol consumption
In the past month, how often did you consume alcohol? Never 2-3 days a month Once a week 2-3 days a week 4-5 days a week 6-7 days a week Approximately how many alcoholic drinks did you consume on the days you drankonly to appear if answer to previous question is 1-2 days or more (one drink is equivalent to 1 beer can (340ml), 1 small glass of wine (150ml), 1 metric tot of spirits (25ml), or 1 small glass of liqueur (25ml))? 0 1 2 3 4 5 or more+ In a typical week, how many times have you had 5 or more drinks on one occasion?
Never 1 time 2-3 times More than 3 times Are you intending to cut down on your alcohol consumption over the next 6 months?only appears if answer to first question is 1-2 days or more, or if answer to previous question is once a week or more

Emotional health and wellbeing
On the whole, people within my organisation get on well with minimal conflict