ORIGINAL ARTICLES

Prevalence of a set of risk factors for chronic diseases in the city of São Paulo, Brazil

Luiz Francisco Marcopito^I; Sérgio São Fins Rodrigues^II; Maria Aparecida Pacheco^II; Mírian Matsura Shirassu^II; Artur Jaques Goldfeder^II; Marco Antonio de Moraes^II

^IDepartamento de Medicina Preventiva. Universidade Federal de São Paulo. São Paulo, SP, Brasil
^IIDivisão de Doenças Crônicas Não-Transmissíveis. Centro de Vigilância Epidemiológica. Secretaria de Estado da Saúde de São Paulo. São Paulo, SP, Brasil

Correspondence

ABSTRACT

OBJECTIVE: To estimate the prevalence of a set of risk factors for non-transmissible chronic diseases and compare it to that found 15-16 years ago in a similar survey.
METHODS: A cross-sectional household survey was carried out comprising a random sample of people aged 15-59 years in the city of São Paulo between 2001 and 2002. The total of 2,103 people answered a questionnaire and had their blood pressure, weight, height, waist and hip circumferences measured. For a third of these participants, their total cholesterol, HDL-cholesterol, triglycerides and glucose levels were determined.
RESULTS: The total age-adjusted prevalences in the study age group were as follows: smoking, 22.6%; uncontrolled blood pressure, 24.3%; obesity, 13.7%; increased waist circumference, 19.7%; total cholesterol >240 mg/dL, 8.1%; HDL-cholesterol <40 mg/dL, 27.1%; triglycerides >200 mg/dL, 14.4%; and blood glucose >110 mg/dL, 6.8%. Smoking, uncontrolled blood pressure, high total cholesterol, low HDL-cholesterol and high triglycerides were significantly more prevalent in men than women.
CONCLUSIONS: The prevalences of a set of risk factors for chronic diseases showed men to have a poorer condition than women. In comparison to the previous survey, the prevalence of uncontrolled blood pressure remained unchanged but the prevalence of smoking has significantly lowered.

Keywords: Chronic disease, epidemiology. Morbidity surveys. Risk factors. Hypertension. Smoking. Obesity. Lipids. Diabetes mellitus. Prevalence.

INTRODUCTION

Chronic non-transmissible diseases comprise a large group of conditions that are by far the leading causes of deaths in the city of São Paulo and in Brazilian urban areas as a whole.^* Individual population-based studies on the prevalence of recognized risk factors for these conditions have been performed in Brazilian communities as large as the city of São Paulo¹⁸ (population of 9.2 million in 1987) and as small as Bambuí⁹ (population of 21,000 in 1996-1997).

Although these studies were carried out with soundmethodology, most of them were conducted separately and atdifferent points in time (Volta Redonda,⁷ 1979-1980;Fortaleza,¹⁹ 1985; Araraquara,¹⁰ 1987;Piracicaba,¹ 1988; Cotia,¹⁵ 1990-1991; Ilhado Governador,⁸ 1991-1992; Pelotas,^6,171992 and 1994; Passo Fundo,²¹ 1995;Catanduva,⁴ 1998), making it difficult to identifygeographical differences and temporal changes. Few exceptions arethe studies on diabetes in Brazilian capitals¹³ and onobesity trends in urban and rural areas.¹⁶ PortoAlegre is perhaps the only Brazilian city where the prevalence ofhypertension was measured at two points in time, in the 70's andin the 90's.⁵

The results on the prevalences of recognized risk factors fornon-transmissible chronic diseases (smoking, excessive alcoholintake, uncontrolled blood pressure, diabetes, poor nutritionalhabits, lipid disorders, obesity, physical inactivity, stress)are inconsistently published, usually making it difficult forexternal comparisons. In this regard, the World HealthOrganization²² has gathered additional informationfrom authors of more recent studies in order to build acomparable international picture on the prevalence of such riskfactors.

The 1987 survey in the city of São Paulo¹⁸ provided statistics on smoking, blood pressure, body mass index, alcohol intake, and physical inactivity at that time. This present study was designed to estimate current statistics for the same risk factors plus others, making comparison when possible. Over those 15-16 years, however, the city has become more populated (10.4 million in 2000) and more violent: the crude death rate for homicides increased 50% from 1987 to 2000.^* As a result, it has become more difficult to carry out household surveys.

METHODS

A cross-sectional household survey was carried out comprisinga random sample of people from four out of six homogeneoussocio-economic areas²⁰ of the city of SãoPaulo. Three districts in each of the four areas were selected,in order to cover similar geographical area as surveyed in1987.¹⁸ The sample size was estimated in 2,100subjects, about 40% larger than that of the 1987 survey.

From the 12 chosen districts 96 census tracts were randomlyselected and checked for the year 2001 and about 25,000households were listed. By systematic sampling a large enoughnumber of household were selected for reaching the estimatedsample size. A total of 2,852 households were visited, of which155 (5.4%) were vacant and 186 (6.5%) were occupied by peopleolder than 59 years.The remaining 2,511 households wereapproached through an invitation letter. If residents agreed toparticipate in the study, a list of all people aged 15-59 yearswas made up by an independent visitor, the list order followingthe pattern recommended by Marques &Berquó,¹⁴ which consists of listing first fromolder to younger males. A total of 401 (16.0%) residents refusedto participate.

In each household one person was selected by a stratifiedrandom process using pre-prepared adhesive labels.¹⁴ Atotal of 2,110 people were home interviewed from March 2001 toOctober 2002, but seven were later excluded because they werefound to be pregnant or puerperal women.

The final sample was then composed of 2,103 people whoanswered a questionnaire applied by trained nurse interviewers,and had their blood pressure, weight, height, waist and hipcircumferences measured twice. Most (70.9%) interviews wereperformed on weekends (42.7% on Saturdays and 28.2% on Sundays),and 29.1% were equally distributed on the other five weekdays. A10% randomly selected sub-sample of interviews had theirintegrity checked by telephone, and showed no discrepancies.

About a third of participants (759), randomly selected, answered a quantitative nutrition questionnaire and 700 had a fasting venous blood sample collected at home; as 59 (7.8%) refused to have their blood drawn. In order to minimize the interference of overeating on weekends, blood collection was not carried out on Sundays and was avoided on Mondays. Blood samples were processed for determinations of plasma glucose and serum total cholesterol, HDL-cholesterol and triglycerides. In order to check for fasting status blood glucose (>126 mg/dl) and triglycerides (>200 mg/dl) were cross-tabulated: only 13 subjects fell in both categories, 11 of whom were known to have diabetes and/or lipid disorder.

Blood pressure was measured using a mercury columnsphygmomanometer in the middle and at the end of questionnaireapplication: systolic and diastolic pressures were those of firstand fifth Korotkoff's phases, respectively. Height was determinedwith the participant barefooted, using a tape measure fixed to awall and a plastic square. Weight was measured using a portabledigital dial scale, and participants wearing only underclothes.Waist and hip circumferences were taken with a tape measure, snugand parallel to the floor. As these measurements were takentwice, the average of both was considered for the analysis.

Participants were classified as smokers if they reported smoking on a daily basis, regardless the amount consumed. Participants with systolic blood pressure >140 mm Hg or diastolic >90 mm Hg were considered as having uncontrolled blood pressure, regardless any other circumstances. The body mass index cut-off point for obesity was >30 kg/m² in both sexes, the cut-off point for increased waist circumference was >102 cm in males and >88 cm females. Prevalences of high serum total cholesterol (>240 mg/dl), low HDL-cholesterol (<40 mg/dl), high triglycerides (>200 mg/dl), and elevated plasma glucose (>110 mg/dl) were determined according to these cut-off points.

For purposes of external comparisons, age was divided into10-year groups, except the youngest (15-29 years). Point estimateprevalences were presented in percentages, by age groups and sexwith their 95% confidence intervals (95% CI), calculated usingthe exact probabilities given by the binomial distribution,because in some cases the normal approximation would nothold.

The crude total prevalences by sex were corrected for thesampling effect using the respective actual age compositions,according to Cochran's⁴ technique for finitepopulations. The 95% CI were calculated using the normalapproximation to the binomial distribution:

95% CI = p ± 1.96 e ,

where e is the sampling error of p, obtained from the square root of the variance (e²) of p, as follows:

where

N_i is the population size in the i^th agegroup,

n_i is the sample size in the i^th agegroup, and

p_i is the sample prevalence in thei^th age group.

This correction generated population estimates of totalprevalences for both men and women. In order to eliminatepossible distortions caused by different age compositions in bothsexes, age-adjusted total prevalences in both sexes werecalculated for comparison purposes, using the age composition oftotal (male and female) target population as standard.

For statistical comparisons confidence intervals were usedrather than p-values. Data entry was duplicated. The electronicprograms SPSS 10.0, Stata 7.0, and Excel 97 were used.

This study was approved by the Research Ethics Committee ofthe Universidade Federal de São Paulo. Participants wereinformed of their measurement results and those with abnormalresults were referred to the nearest health care center.

RESULTS

Table 1 shows the distribution by age and sex of the target population, total sample, and sub-sample. It was observed that, although the total proportion by sex in sample were quite similar to the target population, the sample and sub-sample were somewhat older and had a slightly higher female proportion, justifying corrections and adjustments for both variables.

The prevalences are showed by age groups, allowing for directcomparisons across age groups and in both sexes. The totalprevalences are presented in two ways: those obtained in thesample (crude), and population estimates by sex (individuallycorrected for the actual population age composition of eachsex).

Table 2 shows age-specific and total (crude and corrected) prevalences by sex. Smoking consistently increased with age among men. Across age groups of both sexes, the prevalence of smokers in the 50-59 age group showed to be significantly higher in men. The prevalence of uncontrolled blood pressure increased with age in both sexes, and was statistically higher in males in all age groups. Population estimates of total prevalences of both smoking and uncontrolled blood pressure were statistically higher in men. Concerning obesity, there was a trend of increasing prevalences with age in both sexes, but differences across both female and male age groups were not significant. Increased waist circumference statistically increased with age in both sexes, but the statistical differences between both sexes were probably due to differences in cut-off points (102 cm and 88 cm).

Table 3 shows age group and total (crude and corrected) prevalences by sex obtained from the sub-sample of 700 participants. Population estimate of total prevalence of inadequate total cholesterol was significantly higher in men and it seemed to be more frequent in younger males. Across age groups of both sexes, the prevalence of inadequate total cholesterol in the 40-49 age group was statistically higher in men; the 50-59 age group was the one with the highest prevalence among women. Population estimate of total prevalence of inadequate triglycerides was much higher in men due to increased prevalences in those aged less than 50 years. Although it was observed a trend of increasing prevalence of inadequate plasma glucose with age, there were no statistical differences between both sexes, neither across age groups nor in the population estimates of total prevalences.

Table 4 shows age-specific and total (crude and corrected) prevalences by sex of inadequate HDL-cholesterol in 684 participants who were tested for this substance (16 were excluded due to lipidemia). There was no consistent variation with age in both sexes. Across age groups in both sexes, the extreme age groups were different, and the population estimate of total prevalence was statistically higher in men.

It can be observed that smoking, uncontrolled blood pressure, high total cholesterol, low HDL-cholesterol and high triglycerides remained significantly more prevalent in those men aged 15-59 years, even after adjustment for age (Table 5).

DISCUSSION

The study sample was randomly drawn from a target populationcomprising 88.6% of all São Paulo residents (population of6,869,224) in the 15-59 age group studied. In such a householdsurvey, the refusal percentage (16.0%) was considered acceptablein view of the violence situation in São Paulo, where thecrude death rate for homicides (per 100,000 population) escalatedfrom 38.8 in 1987 to 58.5 in 2000. Fear of violence must havebeen the main reason why people living in more privileged areasrefused to participate. If, on the one hand, this was alimitation in coverage, on the other hand it made the targetpopulation similar to that surveyed in 1987.In general, the studyshowed men to have poorer risk factors for chronic diseases thanwomen across age groups, and there was not a uniform trend ofincreasing prevalence with age.

Comparisons between total prevalences found in this study withthose obtained in the 1987 survey¹⁸ may not becompletely accurate, as the standard age distributions used foradjustments were not the same in both studies. Fully comparableestimates will be showed in forthcoming reports, as the 1987database was fully recovered and is available.

The adjusted prevalence of smoking in this present study wasmuch lower than that found in 1987 survey,¹⁸ both inmales (from 44.6 to 25.4%) and females (from 31.9 to 19.9%).Accordingly, the total smoking prevalence dropped from 37.9 to22.6%. At least three factors may have contributed to thissmoking reduction in the interim of both studies: an actualreduction of population's purchasing power; income reduction dueto increasing unemployment; and the impact of ongoing nationalcampaigns against smoking.

No substantial changes were noted in the prevalences ofuncontrolled blood pressure. Men continued to have a poorercondition (31.0% in 1987, 30.3% in the present study) than women(14.4 and 18.2%, respectively), and the total prevalence did notchange at all (22.3 vs 24.3%). One may only speculateabout this unchanged picture, since hypertension control dependson several factors such as access to health care, diagnosis andtreatment compliance.

Repeated cross-sectional surveys have the disadvantage of showing twice the situations after, rather than during, a certain time period. What is available are two snapshots taken 15 years apart and no information of what happened in between. It is difficult, therefore, to match the declining mortality rates due to stroke^** in São Paulo with the unchanged picture on uncontrolled blood pressure. A similar trend was seen in two surveys carried out 15 years apart in the southern Brazilian city of Porto Alegre.⁵

In regard to obesity, the results are not comparable to thoseof the 1987 survey¹⁸ due to differences in body massindex cut-off points. Waist circumference and blood levels oflipids and glucose were not determined in the 1987 survey.

Still in regard to blood pressure, it must be emphasized thatthe numbers presented here do not refer to the prevalence ofhypertension, as the working definition of hypertension usuallytakes into account the current intake of anti-hypertensive drugs.When there is any degree of control with the use of drugs, theprevalence of hypertension usually is higher than the prevalenceof uncontrolled blood pressure, because drug-controlled peopleare considered to have hypertension.

For purposes of comparison there were selected some Brazilianpopulation-based studies in which working definitions matchedthose applied in this study. Due to differences in the age andsex composition among studies, it was opted for comparingage-specific prevalences in both sexes. When these strata werenot available in the original publications, it was search foradditional information in the WHO Global NCDInfobase.²²

When compared to the prevalences found in the smallcommunities of Ilha do Governador (a district of the city of Riode Janeiro)⁸ and Bambuí (state of MinasGerais),² respectively in 1991-1992 and 1996-1997,age-specific point prevalences of uncontrolled blood pressure aremuch higher (about two to three times) in the city of SãoPaulo. They are even higher than those age-specific pointprevalences of hypertension found in 1998 in the medium-sized(population: 103,000) urban community of Catanduva (state ofSão Paulo).⁴

Age-specific prevalences of smoking observed are much lowerthan those seen in Araraquara (state of SãoPaulo)¹² in 1987. When compared toBambuí,⁹ the prevalence of smoking in men aged50–59 years was the same, but much lower in younger age groups.Among women, the prevalence of smoking in the 50-59 age group istwice as high in the present study, but lower in younger agegroups.

In younger age groups of both sexes, point prevalences ofobesity are similar to those found in 1987 inAraraquara¹¹ and in 1994 in Pelotas (state of RioGrande do Sul),⁶ but in older age groups theprevalences vary according to sex and age. As compared to theresults obtained in a large Brazilian survey^16,22 in1997, the prevalences of obesity are higher in all age groups ofboth sexes, but may not be comparable because the available datarefer to both urban and rural populations.

The prevalences of inadequate total cholesterol in women arevery similar to those obtained in Bambuí.^9,22In men, however, the 40-49 age group shows a prevalence 2.5 timeshigher than that in Bambuí. This may be a feature ofSão Paulo population or may be due to sampling error.

No comparable Brazilian data on the prevalences of inadequateserum triglycerides, plasma glucose and waist circumference areavailable for external comparisons. Result interpretation onincreased waist circumference must take into account thedifferent cut-off points for men and women.

External comparisons with the results of other Brazilianstudies are difficult because the available statistics weregenerated in different urban areas at different points intime.

ACKNOWLEDGEMENTS

To the doctors Ana Maria Sanches and Maria Célia GuerraMedina and nutritionists África Isabel de la Cruz PerezNeumann and Adriana Bouças Ribeiro of the Divisãode Doenças Crônicas Não-Transmissíveisof Centro de Vigilância Epidemiológica of Secretariade Estado da Saúde de São Paulo, for theircriticism and help in organizing and conducting the fieldwork.

REFERENCES

1. Ayres JEM. Prevalência de hipertensão arterial na cidade de Piracicaba. Arq Bras Cardiol 1991;57(1):33-6.        [ Links ]

2. Barreto SM, Passos VMA, Firmo JOA, Guerra HL, Vidigal PG, Lima-Costa MFF. Hypertension and clustering of cardiovascular risk fctors in a community in Southeast Brazil: the Bambuí Health and Ageing Study. Arq Bras Cardiol 2001;77(6):576-81.        [ Links ]

3. Cochran WG. Sampling techniques. New York: John Wiley & Sons; 1953. p. 330.        [ Links ]

4. Freitas OC, Carvalho FR, Neves JM, Veludo PK, Parreira RS, Gonçalves RM, Lima SA, Bestetti RB. Prevalence of hypertension in the urban population of Catanduva, in the State of São Paulo, Brazil. Arq Bras Cardiol 2001;77(1):16-21.        [ Links ]

5. Fuchs FD, Moreira LB, Moraes RS, Bredemeier M, Cardozo SC. Prevalência de hipertensão arterial sistêmica e fatores associados na região urbana de Porto alegre. Estudo de base populacional. Arq Bras Cardiol 1994;63(6):473-9.        [ Links ]

6. Gigante DP, Barros FC, Post CL, Olinto MT. Prevalência de obesidade em adultos e seus fatores de risco. Rev Saúde Pública 1997;31:236-46.        [ Links ]

7. Klein CH, Araújo JWG, Leal MC. Inquérito epidemiológico sobre hipertensão arterial em Volta Redonda, RJ. Cad Saúde Pública 1985;1:58-70.        [ Links ]

8. Klein CH, Souza e Silva NA, Nogueira AR, Bloch KV, Campos LHS. Hipertensão arterial na Ilha do Governador, Brasil. II. Prevalência. Cad Saúde Pública 1995;11:389-94.        [ Links ]

9. Lima-Costa MF, Barreto SM, Uchoa E, Firmo JO, Vidigal PG, Guerra HL. The Bambui Health and Ageing Study (BHAS): prevalence of risk factors and use of preventive health services. Rev Panam Salud Publica 2001;9:219-27.        [ Links ]

10. Lolio CA. Prevalência de hipertensão arterial em Araraquara. Arq Bras Cardiol 1990;55(3):167-73.        [ Links ]

11. Lolio CA, Latorre MRDO. Prevalência de obesidade em localidades do estado de São Paulo, Brasil, 1987. Rev Saúde Pública 1991;25:33-6.        [ Links ]

12. Lolio CA, Souza JMP, Santo AH, Buchalla CM. Prevalência de tabagismo em localidade urbana da região sudeste do Brasil. Rev Saúde Pública 1993;27:262-5.        [ Links ]

13. Malerbi DA, Franco LJ. Multicenter study of the prevalence of diabetes mellitus and impaired glucose tolerance in the urban Brazilian population aged 30-69 yr. The Brazilian Cooperative Group on the Study of Diabetes Prevalence. Diabetes Care 1992;15:1509-16.        [ Links ]

14. Marques RM, Berquó E. Seleção da unidade de informação em estudos de tipo "survey": um método para a construção das tabelas de sorteio. Rev Saúde Pública 1976:10:81-92.        [ Links ]

15. Martins IS, Coelho LT, Mazzilli RN, Singer JM, Souza CU, Antonieto Jr AE et al. Doenças cardiovasculares ateroscleróticas, dislipidemias, hipertensão, obesidade e diabetes melito em população de área metropolitana da região sudeste do Brasil. I- Metodologia da pesquisa. Rev Saúde Pública 1993;27:250-61.        [ Links ]

16. Monteiro CA, Benicio MHDA, Conde WL, Popkin BM. Shifting obesity trends in Brazil. Eur J Clin Nutr 2000;54:342-6.        [ Links ]

17. Piccini RX, Victora CG. Hipertensão arterial sistêmica em área urbana no sul do Brasil: prevalência e fatores de risco. Rev Saúde Pública 1994;28:261-7.        [ Links ]

18. Rego RA, Berardo FAN, Rodrigues SSF, Oliveira ZMA, Oliveira MB, Vasconcellos C et al. Fatores de risco para doenças crônicas não-transmissíveis: inquérito domiciliar no município de São Paulo, SP (Brasil). Metodologia e resultados preliminares. Rev Saúde Pública 1990;24:277-85.        [ Links ]

19. Rouquayrol MZ, Veras FM, Vasconcelos JS, Bezerra RC, Gomes IL, Bezerra FA. Fatores de risco na doença coronária. Inquérito epidemiológico em estratos habitacionais de um bairro de Fortaleza. Arq Bras Cardiol 1987;49(6):339-47.        [ Links ]

20. Sposati A, coordenador. Mapa da inclusão/exclusão social da cidade de São Paulo: Editora da PUC-SP; 1996. p. 124.        [ Links ]

21. Trindade IS, Heineck G, Machado JR, Ayzemberg H, Formighieri M, Crestani M, Gusso J. Prevalência da hipertensão arterial sistêmica na população urbana de Passo Fundo (RS). Arq Bras Cardiol 1998;71(2):127-30.        [ Links ]

22. WHO Global NCD Infobase. Available from: URL: http://www.who.int/ncd_ surveillance/infobase/en/ [2003 Apr]        [ Links ]

Correspondence to
Luiz Francisco Marcopito
Departamento de Medicina Preventiva - Unifesp
Rua Borges Lagoa, 1341
04038-034 São Paulo, SP, Brasil
E-mail: marcopito@medprev.epm.br

Received on 26/2/2004. Reviewed on 31/5/2005. Approved on 17/6/2005.
Funded by the Fundação de Amparo à Pesquisa do Estado de São Paulo (Fapesp - Grant n. 99/06187-6)

* Datasus. Available from URL: http://tabnet.datasus.gov.br [fev 2004]
** Centro de Vigilância Epidemiológica, Estado de São Paulo. Available from: URL: http:// www.cve.saude.sp.gov.br [fev 2004]