Identification of dietary patterns using factor analysis in an epidemiological study in São Paulo

",
É
,/  É $IRCE -ARIA ,OBO -ARCHIONI -ARIA DO 2OSÉRIO $IAS DE /LIVEIRA ,ATORRE *OSÏ %LUF .ETO 6ICTOR 7àNSCH &ILHO 2EGINA -ARA &ISBERG )DENTIlCATION OF DIETARY PATTERNS USING FACTOR ANALYSIS IN AN EPIDEMIOLOGICAL STUDY IN 3ÎO 0AULO &ACULDADE DE 3A¢DE 0¢BLICA 5NIVERSIDADE DE 3áO 0AULO 3áO 0AULO "RAZIL
-/,
/ #/.4%84 !.$ /"*%#4)6% $IET AND NUTRITION ARE ENVIRONMENTAL FACTORS IN HEALTHDISEASE RELATION SHIPS &ROM THE EPIDEMIOLOGICAL VIEWPOINT DIET REPRESENTS A COMPLEX SET OF HIGHLY CORRELATED EXPOSURES /UR OBJECTIVE WAS TO IDENTIFY PATTERNS OF FOOD INTAKE IN A GROUP OF INDIVIDUALS LIVING IN 3ÎO 0AULO AND TO DEVELOP OBJECTIVE DIETARY MEASUREMENTS FOR EPIDEMIOLOGICAL PURPOSES $%3)'. !.$ ,/#!, %XPLORATORY FACTOR ANALYSIS OF DATA IN A CASE CONTROL STUDY IN SEVEN TEACHING HOSPITALS IN 3ÎO 0AULO -%4(/$34HE PARTICIPANTS WERE PATIENTS  ORAL CANCER CASES AND  CONTROLS ADMIT TED TO THE STUDY HOSPITALS BETWEEN .OVEMBER  AND -ARCH  4HE WEEKLY INTAKE FREQUENCIES FOR DAIRY PRODUCTS CEREALS MEAT PROCESSED MEAT VEGETABLES PULSES FRUITS AND SWEETS WERE ASSESSED BY MEANS OF A SEMI QUAN TITATIVE FOOD FREQUENCY QUESTIONNAIRE $IETARY PATTERNS WERE IDENTIlED BY FACTOR ANALYSIS BASED ON THE INTAKE OF THE EIGHT FOOD GROUPS USING PRINCIPAL COMPONENT ANALYSIS AS AN EXTRACTION METHOD FOLLOWED BY VARIMAX ROTATION 2%35,43 &ACTOR ANALYSIS IDENTIlED THREE PATTERNS THAT ACCOUNTED FOR  OF THE TOTAL VARIABILITY WITHIN THE SAMPLE 4HE lRST PATTERN hPRUDENTv WAS CHARACTERIZED BY VEGETABLE FRUIT AND MEAT INTAKE THE SECOND hTRADITIONALv BY CEREALS MAINLY RICE AND PULSES MAINLY BEANS  AND THE THIRD hSNACKSv BY DAIRY PRODUCTS AND PROCESSED MEAT #/.#,53)/. 4HIS STUDY IDENTIlED FOOD INTAKE PATTERNS THROUGH AN A POSTERIORI APPROACH 3UCH ANALYSIS MAY BE USEFUL FOR NUTRITIONAL INTERVENTION PROGRAMS AND AFTER COMPUTING SCORES FOR EACH INDIVIDUAL ACCORDING TO THE PATTERNS IDENTIlED FOR ESTABLISHING A RELATION SHIP BETWEEN DIET AND OTHER EPIDEMIOLOGICAL MEASUREMENTS OF INTEREST +%9 7/2$3 &OOD HABITS $IET $IET SURVEYS &EEDING BEHAVIOR .UTRITIONAL ASSESSMENT  /," 1 /" $IET AND NUTRITION ARE IMPORTANT FACTORS IN THE PROMOTION AND MAINTENANCE OF GOOD HEALTH THROUGHOUT THE ENTIRE COURSE OF LIFE &OR A LONG PERIOD OF TIME THE MAJOR CONCERNS OF RESEARCHERS AND HEALTH PROFESSIONALS WERE IN RELATION ONLY TO THE PREVENTION OF DElCIENCY RELATED DISEASES SUCH AS SCURVY AND BERIBERI /VER RECENT DECADES HOW EVER THERE HAS BEEN ADDITIONAL CONCERN RELATING TO THE PREVENTION OF CHRONIC NON COMMUNICABLE DISEASES SUCH AS CARDIOVASCULAR DISEASES CANCER DIABETES HYPERTENSION AND OSTEOPOROSIS 4HE COMPLEXITY OF THE HUMAN DIET PRESENTS A CHALLENGE TO THOSE INTENDING TO STUDY THE RELATION SHIP BETWEEN DIET AND DISEASE $IET HAS TRADITION ALLY BEEN STUDIED IN TERMS OF NUTRIENTS (OWEVER IT SHOULD BE EMPHASIZED THAT FOODS CONTAIN OTHER CHEMICAL COMPOUNDS SOME WELL KNOWN SOME STILL POORLY CHARACTERIZED AND OTHERS COMPLETELY UNKNOWN AND WHICH AT PRESENT CANNOT BE MEA SURED &URTHERMORE THE DIVERSITY OF FOOD COM BINATIONS MAY LEAD TO COMPETITION ANTAGONISM OR ALTERATION IN NUTRIENT BIOAVAILABILITY &ROM AN EPIDEMIOLOGICAL VIEWPOINT THE DIET REPRESENTS A COMPLEX SET OF HIGHLY CORRELATED EXPOSURES4HUS THE REAL RELATIONSHIP BETWEEN A FOOD GROUP AND A DISEASE MAY ERRONEOUSLY BE ATTRIBUTED TO A SINGLE COMPONENT BECAUSE OF THE MULTICOLLINEARITY THAT EXISTS BETWEEN NUTRIENTS AND FOODS  /NE OPTION WHEN DEALING WITH THE COMPLEX ITY OF INTERCORRELATIONS BETWEEN FOODS IS THE USE OF PATTERN ANALYSIS 4HIS APPROACH USES THE CORRELA TIONS BETWEEN FOOD AND NUTRIENT INTAKE TO DESCRIBE A GENERAL DIETARY PATTERN THAT AT A LATER STAGE MAY BE RELATED TO THE RISK OF A DISEASE 4HIS APPROACH IS OF PARTICULAR VALUE IF THE EFFECT OF THE DIET IS NOT MEDIATED BY ONE OR TWO SPECIlC NUTRIENTS BUT BY NUTRIENTS THAT PERHAPS OPERATE INTERACTIVELY   4WO APPROACHES HAVE BEEN USED FOR DEVELOP ING GENERAL DESCRIPTORS OF DIETARY PATTERNS 4HE lRST APPROACH CALLED A PRIORI IS BASED ON PREVIOUS KNOWLEDGE OF THE FAVORABLE AND UNFAVORABLE EF FECTS OF DIET CONSTITUENTS EG BY USING THE $IET 1UALITY )NDEX  !NOTHER APPROACH A POSTERIORI IS BASED ON THE DIETARY DATA OBTAINED 4HE MAIN TECHNIQUES IN THIS LATTER APPROACH ARE PRINCIPAL COMPONENT ANALYSIS FOLLOWED BY FACTOR ANALYSIS AND THIS APPROACH REQUIRES STATISTICAL MODELING 4HE GOAL IS TO TRANSFORM A LARGE SET OF CORRELATED VARIABLES INTO A SMALLER SET OF NON CORRELATED VARIABLES CALLED PRINCIPAL COMPONENTS OR FACTORS )N FACTOR ANALYSIS RATHER THAN ESTABLISHING A DIET INDICATOR THE DATA OBJECTIVELY POINT TOWARDS HOW MEASUREMENTS ARE CLUSTERED 4HE AIM OF THIS TECHNIQUE IS TO IDENTIFY THE UNDERLYING STRUCTURE IN A DATA MATRIX BY SUMMARIZING AND REDUCING DATA IN ORDER TO SUPPLY A SYNTHETIC MEASUREMENT OF THE DIET )N ORDER TO SUMMARIZE THE DATA FACTOR ANALYSIS DERIVES DIMENSIONS THAT WHEN INTERPRETED AND UNDERSTOOD DESCRIBE THE DATA IN TERMS OF A MUCH SMALLER NUMBER OF ITEMS THAN DO THE INDIVIDUAL VARIABLES  4HE AIM OF THE PRESENT STUDY WAS TO DE SCRIBE THE FOOD PATTERNS IN A GROUP OF INDIVIDU ALS WHO WERE PARTICIPATING IN A ,ATIN !MERICAN CASE CONTROL STUDY AND TO RELATE ENVIRONMENTAL FACTORS TO CANCER OF THE ORAL CAVITY AND LARYNX THROUGH THE A POSTERIORI APPROACH  /" 4HE PRESENT STUDY UTILIZED DATA OBTAINED FROM A MULTICENTER HOSPITAL BASED CASE CONTROL STUDY THAT WAS PERFORMED WITH THE SUPPORT OF THE )NTERNATIONAL !GENCY FOR 2ESEARCH ON #ANCER )!2#  "ETWEEN .OVEMBER  AND -ARCH   PATIENTS WERE RECRUITED INCLUDING  CASES OF CANCER OF THE ORAL CAVITY OROPHARYNX AND HYPOPHARYNX THAT HAD BEEN IDENTIlED IN SEVEN HOSPITALS IN THE MUNICIPALITY OF 3áO 0AULO AND  CONTROLS THAT HAD BEEN ADMITTED TO THESE SAME HOSPITALS BECAUSE OF CONDITIONS THAT WERE UNRELATED TO DISEASES AS SOCIATED WITH RISK FACTORS FOR CANCER OF THE ORAL CAVITY 4HE STUDY WAS APPROVED BY THE .ATIONAL #OMMITTEE FOR %THICS IN 2ESEARCH #OMISSáO .ACIONAL DE ›TICA EM 0ESQUISA #/.%0  3AO 0AULO -ED *    '(+ $IETARY INTAKE INFORMATION WAS COLLECTED USING A SEMIQUANTITATIVE FOOD FREQUENCY QUES TIONNAIRE &&1  4HE &&1 LIST CONSISTED OF  FOODS FOOD GROUPS OR PREPARATIONS %ACH PARTICIPANT IN THE STUDY WAS ASKED TO PROVIDE FOR EACH ITEM OF THE &&1 THE MEAN WEEKLY INTAKE FREQUENCY BEFORE THE EMERGENCE OF DIS EASE SYMPTOMS 4HE ANSWERS WERE OPEN THUS ALLOWING THIS VARIABLE TO BE TREATED AS CONTINU OUS 4HE FOODS ON THE QUESTIONNAIRE WERE THEN CLASSIlED INTO THE FOLLOWING FOOD GROUPS DAIRY PRODUCTS MILK YOGURT AND CHEESE  CEREALS BREAD RICE PASTA AND MAIZE  MEAT BEEF PORK POULTRY AND lSH  PROCESSED MEAT SAUSAGE SALAMI AND HAM  VEGETABLES RAW VEGETABLES CRUCIFEROUS PLANTS TOMATOES AND CARROTS  PULSES BEANS AND PEAS  FRUITS APPLES PEARS BANANAS AND FRUIT JUICES AND SWEETS DESERTS SWEETS AND CAKES  4HESE GROUPINGS WERE BASED ON THE SIMILARITY OF NUTRIENT CONTENT -Ë>ˇÂˇV>ŽÉ>—>ŽÜ‡ $IETARY PATTERNS WERE OBTAINED BY EXPLOR ATORY FACTOR ANALYSI (...truncated)