Economc crss and follow-up of the condtons that defne metabolc syndrome n a cohort of Catalona, 2005-2012 Laa Maynou 1,2,3, Joan Gl 4, Gabrel Coll-de-Tuero 5,2, Ton Mora 6, Carme Saurna 1,2, Anton Scras 7, Marc Saez 1,2,8 XXXIII Jornadas AES, Santander, June, Thursday 20, 2013 1 Research Group on Statstcs, Econometrcs and Health (GRECS), Unversty of Grona; 2 CIBER of Epdemology and Publc Health (CIBERESP); 3 Autonomous Unversty of Barcelona; 4 Unversty of Barcelona; 5 Insttute of Health Care (IAS), Grona; 6 Unverstat Internaconal de Catalunya; 7 Badalona Serves Assstencals; 8 Center for Research n Health and Economcs (CRES), Unverstat Pompeu Fabra, Barcelona
Introducton Followng Dávla-Quntana and Gonzalez Lopez-Valcárcel (2009) s reasonable to assume that, f any, the effects of the recesson on health would be, n any case, ndrect, short term and reversble. The problem s that there s lttle evdence on the mpact of the recesson on health. Dávla-Quntana CD, González-López-Valcárcel. Economc crss and health [n Spansh]. Gaceta Santara 2009; 23(4):261-265.
Introducton Among the effects of the recesson, several authors have argued that there mght have been a change n detary patterns. To the extent that, as a consequence of the recesson, famles are mpovershed and, on the other hand, the relatve prces of healthy food have ncreased, mght expect that those famles opted to ncrease ther consumpton of (cheaper) unhealthy food (.e. junk food). However there s no quanttatve evdence and the conclusons of the most recent papers are nothng more than workng hypotheses that deserve further research. - Drewnowsk A, Specter SE. Poverty and obesty: the role of energy densty and energy costs. Am J Cln Nutr. 2004;79:6-16. - Darmon N, Drewnowsk A. Does socal class predct det qualty? Am J Cln Nutr. 2008;87:1107-1117. - Suhrcke M, Stuckler D. W ll the recesson be bad for our health? It depends. Socal Scence & Medcne 2012; 74:647-653. - Wnters L, McAtteer S, Scott-Samuel A. Assessng the mpact of the economc downturn on health and wellbeng. Lverpool Publc Health Observatory, Observatory Report Seres No. 88, February 2012.
Introducton In an attempt to broaden the evdence, we proceeded to conduct a lterature revew, We found twenty-one ctatons. Most of them, however, contans narratve revew, merely dscussed hypothess or proposed future research lnes. Only egth artcles provde quanttatve evdence regardng the current recesson. Only Powell and Chaloupka (2010), through a systematc revew of studes between 2008 and 2010, found that small taxes or subsdes, alterng the cost of unhealthy (energy-dense foods) compared wth healthy (less-dense foods) are not lkely to produce sgnfcant changes n body mass ndex or obesty prevalence (although t may n some groups). Powell LM, Chaloupka FJ. Food prces and obesty: evdence and polcy mplcatons for taxes and subsdes. Mlbank Q 2010;87(1):229-257.
Introducton We could summarze very lttle evdence, sayng that, as a result of the economc recesson, the ncrease n the relatve prces of healthy foods had ncreased the consumpton of unhealthy foods. As a consequence, there may have been an ncrease n the ncdence of obesty and/or dyslpdema, at least n that most mpovershed. Obesty and dyslpdema are two of the condtons that compose the metabolc syndrome. Ths s defned as the fulfllment of the Natonal Cholesterol Educaton Program Adult Treatment Panel III (ATP-III) fve crtera of dagnoses: abdomnal obesty (wast greater than 102 cm n men and 88cm n women); hypertenson ( 130/85 mmhg); low hgh densty lpoprotens (HDL) (less than 40mg/dL n men and 50mg/dL n women) and hypertrglycerdema ( 150 mg/dl) (these latter two crtera defnng dyslpdema); and glucose ntolerance ( 110 mg/dl or known dabetes).
Objectve Our objectve n ths paper was to analyze the ncdence of the four condtons defnng the metabolc syndrome, namely, obesty, dyslpdema, hypertenson and dabetes melltus type II, by followng a general populaton cohort from 2005-2012. We were nterested, n partcular, n the evoluton of the ncdence durng the economc crss perod,.e. 2009-2012. Our hypothess s that nputs to the condton of metabolc syndrome have exceeded outputs from 2009 onwards.
Methods We used a general populaton retrospectve cohort, composed of ndvduals assgned to one of three Health Basc Areas (ABS, Àrea Bàsca de Salut ) managed by the Insttute of Health Care (IAS, Insttut d Assstènca Santàra ), whch had some contact wth the health servces between January 1, 2005 and December 31, 2012. We excluded ndvduals under 15 years (medan of 36477 persons per year; Q1: 33025; Q3: 37864). The IAS managed all the ABS that provde health care to the regon of La Selva Interor, Grona (ABS Anglès; ABS Breda-Hostalrc; and ABS Cassà de la Selva). They are manly rural (or sem-urban), wth many towns scattered and dspersed wth farms, estates and small vllages far away.
Methods - Selecton bas-free estmaton The cohort we used was not a random sample of the populaton. In fact, there was selecton bas. Some ndvduals could have a hgher probablty of havng contacted wth the health servces and, therefore, mght have been overrepresented n the sample observed. To estmate the ncdence we used a selecton bas-free method proposed by Saez et al. (2009). The method requres the use of a populaton health survey. In ths paper we chose to use the ESCA 2006, carred out between December 2005 and July 2006. Saez M, Barceló MA, Coll de Tuero G. A selecton-bas free method to estmate the prevalence of hypertenson from an admnstratve prmary health care database n the Grona Health Regon, Span. Computer Methods and Programs n Bomedcne 2009; 93:228-240.
Methods - Selecton bas-free estmaton In the frst stage, we proceed as f the selecton were exogenous and we weghted the sample n such a way as to award less weght to those ndvduals wth greater probablty of beng observed. o The weghtng use was equal to the nverse percentage represented by the users over the populaton (for 2005-2006) of the muncpalty where the user lved. The problem, however, was that the selecton was endogenous, meanng that the unobserved factors whch would nfluence the use of prmary health care would not be ndependent of those non-observables whch affect the response. Second stage. Usng the weghtng sample obtaned n the frst stage, we re-weghted the sample accordng to the probabltes of the use of prmary health care. o These probabltes were estmated usng a two-part econometrc model, termed hurdle model.
Methods Hurdle model The frst part of the decson process (the decson to seek care ) was modelled usng a bnomal lnk logstc- (response: use 12 months ESCA 2006) 1, 1, log 1, 1 1, Prob Y 1 1, Var Y 1 1 In the second part (the frequency of vsts) the dstrbuton of use (condtonal to some use) was modelled as a truncated negatve bnomal (response: number of vsts 2005-2006 cohort) 1, 1, f Y 2 Y, Y 0 2, Y 1 2 2, 2, 2 Yt 2, 2 2 2 1 1 2, exp 2, 1 2 2,
Methods Lnear predctor t 6 0 1Sex k Age Groupk, 7HTA 8DM2 9OBES 10 k2 CHOL Mxed model: We allowed the ntercept, 0, to be random effects,.e. to be dfferent for each ndvdual, capturng characterstcs ndvdual specfc (.e. ndvdual heterogenety). Note that as covarates we ncluded need varables (obesty, hypertenson, dabetes melltus type II and hypercholesterolema) that could condton the use of health servces, and varables, such as sex and age group, modulatng the effect of the need varables.
Results Prevalence of chronc condtons components of metabolc syndrome 2006 2011 2012 ESCA 1 ENSE 3,4 Cohort ESCA 1 ENSE 3,4 Cohort Cohort HTA 15 years 19.8% 20.6% 14.4% 25.2% 25.7% 21.1% 21.5% Males 18.3% 18.6% 13.1% 24.5% 23.5% 20.5% 21.0% Females 21.2% 22.6% 15.8% 25.9% 27.5% 21.7% 22.2% DM2 15 years 5.1% 6.2% 4.1% 8.0% 8.8% 6.8% 7.3% Males 4.6% 6.3% 4.4% 8.0% 8.7% 7.6% 8.1% Females 5.5% 6.0% 3.7% 8.1% 8.9% 6.1% 6.5% Obesty 2 15 years 11.2% 15.7% 6.4% 11.8% 17.8% 14.0% 15.6% Males 11.2% 16.2% 4.0% 12.6% 18.7% 11.1% 12.7% Females 11.3% 15.3% 8.7% 11.0% 17.0% 16.9% 18.4% Hgh cholesterol 15 years 15.0% 15.7% 12.5% 21.9% 21.6% 22.2% 25.2% Males 14.5% 15.3% 12.1% 21.2% 20.6% 22.6% 24.9% Females 15.4% 16.1% 12.8% 22.7% 22.5% 23.1% 25.4% Metabolc syndrome 15 years 21.32% 30.34% 32.35% Males 17.58% 26.79% 30.19% Females 23.05% 33.37% 36.66% Weghted usng the elevaton factors of ESCA (Catalan Health Survey) and ENSE (Spansh Natonal Health Survey) and the Hurdle model for the cohort 1 Self-declared 2 Weght and Heght self-declared n ESCA and ENSE 3 Includng Catalona 4 Declares that the doctor has sad that suffers the condton
Results Incdence of obesty
Results Incdence of hgh cholesterol
Results Incdence of obesty and hgh cholesterol
Results Incdence of hypertenson
Results Incdence of dabetes melltus type II
Centro de de Investgacón Bomédca en en Red Epdemología y y Salud Públca Conclusons There s lttle evdence on the mpact of the recesson on health. As a result of the economc recesson, the ncrease n the relatve prces of healthy foods had ncreased the consumpton of unhealthy foods. As a consequence, there may have been an ncrease n the ncdence of obesty and/or dyslpdema, at least n that most mpovershed. Usng a method that avods the selecton-bas (consequence of usng a non-random cohort) we found that, after the economc recesson (2009-2012), there are: o An mportant ncrease n the prevalence of obesty, hgh cholesterol, hypertenson and dabetes melltus type II. o A sgnfcant ncrease n the ncdence of hgh cholesterol and, n a lesser extent, hgh cholesterol.
marc.saez@udg.edu http://www.udg.edu/grecs.htm