DIET SURVEY INTAKES (PER REFERENCE MAN, FROM THREE-DAY WEIGHED HOUSEHOLD SURVEY)

Transcription

1 477 - FR 333 F H 639 R R (R RFR, FR HR- H HH R) (, ) ist of diet survey intake () variables displayed (abbreviated names) () ( ) 18 macronutrients F 3 R 4 R 5 %F 7 %R 8 %R 9 %R 28 F 29 F 31 %F 32 %F 33 R 34 R 35 %R 36 %R 53 F 59 F 16 micronutrients R H 16 RF a 19 Fe 2 u g 23 n 24 a 26 err 27 Zn 14 staple foods R 38 HFR 39 HR 4 HR H 43 R FH HR 5 amino acids H+ 78 HR 79 RH 23 fatty acids F 83 F 84 F 85 H 86 /R 87 %F 88 %F 89 %F 9 / 91 / 92 n3 93 n6 94 n9 95 %n3 96 %n6 97 %n9 136 %14: 14 %16: 141 %16:1 145 %18: 146 %18:1 147 %18:2 148 %18:3 R : : 11-12

2 478 FR FR R R dentifier: 5 = iet survey variable 5 5 = 5 hort title Full title, including units 5 %F diet survey R F R FR F (for reference man) haded bullets indicate low, middle and high ranges in study areas. n most cases, each category includes about 1/3 of the mainland counties. comparable dietary survey was not carried out in aiwan. 1/ H R F /4 69% 29:F 73% 49: 67% 175:d H H 55% 2:HH 42% 3:H 6% 1:H ap showing all of hina, with study counties designated by shaded bullets :%F 8:%R 9:%R99% %R %R 46% %F -59% -97% 7-33% 18% 4 larke et al., % % ean county values (male and female, all ages), 1983 vs ach county is represented by the letter designating its province (so where there is more than one study county per province, letters are repeated). oints above the diagonal indicate temporal increases in measured values X ( ) H R X F X X ( ) ean small-area values in 1989, xiang vs. xiang. ndicator of reliability of county mean values (i.e., xiangs and combined). 1989

3 479 orrelations of county-specific values, xiang vs. xiang (1989) and, where values for both periods are available, 1983 vs (with both areas combined). = number of counties contributing to each correlation; r% = correlation coefficient: strong correlations indicate informative data. ( ( = r% = oastal and inland counties are grouped and their means per reference man compared statistically. 5 %F ( ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 13.7 R H F F ean ( ) 16.5* 2.5* (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) H -31 * 12 R 38 * a * 14 Rmn F 3 5 c%h2-32 * mn F c% -3 8 c R 9 2 F d -34 * 18 HRc 31 * 1 -R R * 68 df Hc R 68 7 %R F 37 * 94 dh c -36 * 15 -H %R F 31 * 97 drhr Hc H %R F dr c -36 * 19 -R 53 1 R H dr Rc F -36 * 15 H /R dh 39 * 35 mc 42 3 e Fe %F dz Rc FRR u -37 * 88 %F d -33 * 38 RXc RF / d Rc g 32 * 91 / 34 * 164 df 3 42 c m F n df c >2m F 5 93 n dfh Rc -27 R2 RFF %F n df c -36 * R9 14: %F %n df Rc -4 R1 16: R -38 * 96 %n dr c 5 R14 24: R %n dfr RRHb -45 R15 16:1n %R : dfh RRHc 3 R21 2:5n %R %14: d Rc 28 R26 2:4n6 38 * 37 R %16: d 35 * 97 Rb l/cre HFR %18: d 36 * 98 Rc -34 * 2 /cre HR %18: e 34 * 17 a -6 3 a/cre HR -38 * 147 %18:2-31 * 247 fadj Ha R/cre %18: * 119 Ra -38 * 7 R/cre 73 5 R arr -34 * H 42 9 R/cre R dr R 55 2 HH /cre FH er 39 5 H ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. ogether, %fat (5:%F), % plant protein (8:%R) and % carbohydrate (9:%R) make up 99% of dietary calories, so increases in any of these three percentages must imply decreases in another. espite this, %R is positively correlated (46% ) with %R, so both of them are strongly negatively correlated (-59% and -97%, respectively) with %F. Higher along the coast and in the northern herding areas. ide range of values (7-33%; mean equals 18%) that are all lower than the mean of about 4% (larke et al., 1997). he wide range of values is reliable, as there is a very good correlation between xiangs (81%, and a good correlation with 1983 values (55% ). he very low values of about 1% in the three counties in Henan (province ) are reliable, as they occur in both xiangs (and were also seen in the 1983 survey). he relatively high values of about 3% in some of the counties in uangdong (province ) are reliable as they were seen in both xiangs but represent a real increase since ncrease by one quarter in overall mean between 1983 and 1989 surveys probably represents a real increase in percentage fat in the diet. ositively correlated with variables related to animal food intake (e.g., 69% 29:F; 73% 49:; 67% 175:d) and with H, nonh and total cholesterol (55% 2:HH; 42% 3:H; 6% 1:H). his variable correlates with so many factors that causal relationships are difficult to assess. R otes and comments orrelation coefficients (mainland only), p-values and names for all variables with a 2p<.5 correlation with the variable featured on this double page. here are 392 eligible variables (printed in non-italics in the ummary tatistics on pages 19-13), including the 333 with 2-page displays. 2p<.5 p

4 F H R X X X ( ) X ( ) F H R diet survey R (kcal/day/reference man) 48

5 481 1 ( / / ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 2412 R H F F ean ( ) 2499* 2292* (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) 3 19 RHb 3 R4 F %F n6 3 2 RHc -27 R5 n6 31 * R %16: b -29 R7 F RF 3 35 %R -35 * 18 ah Rb -35 * R14 24: %R c% b 27 R18 22:1n9 37 * 18 a 38 * * 96 dr 33 * 69 Rc 34 * R19 24:1n Fe F dfadj 35 * 72 c -27 R26 2:4n6 5 2 u 34 * dfh -31 * 1 H 24 4 a/cre 34 * * 218 eh1st H R/cre g H+ -33 * 219 eh2nd F n HR fadj 35 * R 3 24 a RH 35 * 247 fadj -32 * 13 R 88 4 R a 4 83 F H -34 * 3 e %R Zn -36 * 86 /R Zn -34 * 7 %R F 25 9 / m 25 9 %R %F 31 * 92 n3 ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. 99% of dietary calories are from carbohydrate (mean 71%; see 9:%R), fat (mean 18%; see 5:%F) and plant protein (mean 9%; see 8:R), and only about 1% from animal protein (see 7:%R). wofold range of county values (15-3 kcal/day) is driven chiefly by variation in carbohydrate intake (4:R), so there is a close (88% ) correlation with this variable. ood correlation between xiangs (65% ) but modest correlation with 1983 values (46%*). onsistency between xiangs at upper and lower end confirms reliability of twofold range of county values. eak negative correlations with plasma lipids (e.g., -31%*; 1:H). ean (24 kcal/day) is similar to that in middle-aged males, but the distribution of fat and protein in total calories is much lower than in the % 71% 9:%R 18% 5:%F 9% 8:R 1 7:%R kcal/ 4:R 88% 65% %* 2-31%* 1:H 24 kcal/ R : : 11-12

6 482 2 F diet survey (g/day/reference man) H R F gm/ 4 7 / % % 68% 5:R 77% 55:F 54% 175:d43% 1:H 48% 2:HH 54% 5: 5% 31:a 46% 19:dR X ( ) H X R F X X ( )

7 483 2 F ( / / ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 36.5 R H F F ean ( ) (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) HH %R F 5 31 a c 26 3 H %R -32 * c% -33 * 18 HRc R d 32 * 21 Hc HR dh c 3 9 -R -33 * 15 H 33 * 79 RH 3 95 dh Hc H RF F 4 97 drhr Rc u F drrh 36 * 35 mc -34 * 19 -R F 35 * 149 dr -35 * 38 RXc F g H dr Rc 26 3 e a /R dh 26 4 Hc 32 * F 36 * 87 %F dz c 3 41 m %F %F d c -3 R9 14: 6 32 %F / d Rb -32 * R1 16: -35 * 33 R n d Rc 32 * R14 24: R n6 39 * 164 df c -48 R15 16:1n %R n df Rc -28 R25 2:3n %R %n3 38 * 166 dfh c l/cre R %n6 37 * 167 df RRHc -31 * 2 /cre -37 * 38 HFR 33 * 97 %n9 31 * 169 df c -5 3 a/cre HR : dr c R/cre HR %16: dfr 3 96 Rc R/cre R 34 * 145 %18: dfh Rb /cre * 146 %18: d Rc -32 * 12 R %18:2 35 * 176 d a -31 * 14 Rmn 68 5 R %18: e Ha mn R arr Ra FH ah R Ra 9 5 %F dr 27 5 H 43 1 H 5 7 %R F er ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. Fourfold range of values (2-8 gm/day), with higher values in coastal provinces and in northwest herding areas. onsistency between xiangs at upper and lower end confirms fourfold range of county values. ow values are extremely low in comparison with estern diets. ean increase (about 7 gm/day) probably indicates a real increase in fat consumption since ood correlation between xiangs (74% ) and with 1983 values (49% ). trong correlations with animal food and fat consumption (e.g., 68% 5:R; 77% 55:F; 54% 175:d), plasma lipids (43% 1:H, 48% 2:HH, 54% 5:), and some measures of prosperity and education (e.g., 5% 31:a, 46% 19:dR). R : : 11-12

8 484 3 R diet survey R (g/day/reference man) H R F X ( ) X R X F H X ( )

9 485 3 R ( / / ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 58.1 R H F F ean ( ) (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) -34 * 12 Fc H 5 8 %R H d c -36 * 24 F HR d -36 * 22 b H 7 79 RH d c RF 3 83 F 37 * 112 dfadj -37 * 3 Rb m /R df b -36 * 42 Hsg Fe %F d<25f b Hsb 72 2 u %F dr RRHb >2f * 9 / dh Rb 28 R9 14: g / df Rc 26 R11 18: n 32 * 92 n dr -38 * 97 Rb -24 R16 18:1n err n d c 26 R2 18:3n3 38 * 27 Zn %n3 32 * 177 d a -25 R26 2:4n F %n e%fr -34 * 117 a 25 1 l/cre R %n fadj Ra 29 3 a/cre R %18: 31 * 245 fhadj * 5 /cre HFR %18: fadj R 55 6 R/cre R %18: R -35 * 23 3mn %18: Z-R a F 41 9 dhh 41 4 R H F dh H 32 * R d HF 51 6 %R 34 * drhr ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. ntake higher in the north, and therefore correlated with other factors that show strong north-south variation. wofold range of county values (4-8 g/day). ood correlation between xiangs (6% ) and with 1983 values (47% ). For subdivision of protein intake into plant and animal protein, see 33:R and 34:R, 35:%R (mean % total protein equals 88%) and its complement 36:%R (mean % total protein equals 12%). orrelated with consumption of wheat and other plant food, which is the main source of dietary protein; on average, 8-9% of total protein comes from plant food (see mean values for 33:R). orrelations include: 58% 38:HFR, 39% 88:%F, 51% 158:dH) / 6% % 33:R 34:R, 35:%R 88% 36:%R 12%, 8-9% 33:R 58% 38:HFR 39% 88:%F 51% 158:dH R : : 11-12

10 F H R X X X ( ) X ( ) F H R R diet survey RHR (nitrogen free extract) (g/day/reference man) 486

11 487 4 R ( ) ( / / ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 463 R H F F ean ( ) (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) RHb 32 * 47 >2m R R %18: 27 2 RHc 29 R2 RFF 34 * FH %18: b 25 R4 F H * 147 %18: c -29 R5 n a 34 * 59 F dr Rb -28 R7 F Fe 38 * * 31 a Rc -26 R8 / 6 2 u H+ -32 * 5 c%h Rc 29 R1 16: * 78 HR df 36 * 56 b -51 R14 24: g RH dh 35 * 69 Rc 27 R19 24:1n n F 31 * 96 dr 35 * 72 c -28 R21 2:5n Zn F dh H -31 * R26 2:4n F H -36 * 166 dfh HH 45 1 l/cre F /R df H 43 3 a/cre %F %F dr a/cre %F 32 * 88 %F dfr R/cre R %F dfh 36 * R/cre R 33 * 9 / d -33 * 9 -R %R / d -37 * 13 R 43 3 R %R n9-31 * 218 eh1st H %F HFR 35 * 96 %n eh2nd e %R %n fadj -31 * 33 FRR %R : * 41 m 67 9 %R * 141 %16:1-35 * 5 H ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. ood correlation between xiangs (72% ) and with 1983 values (59% ). ntake higher in the north. ominated by consumption of plant food (7% 28:F). trongly negatively correlated with animal protein consumption in the 3-day diet survey (-53% 34:R), with days per year eating fish or meat reported in the questionnaire (-51% 174:FH, -48% 175:), and with plasma cholesterol (-54% 1:H), providing validation of the diet survey, questionnaire and blood biochemistry % % 7% 28:F 3-53% 34:R -51% 174:FH -48% 175: -54% 1:H R : : 11-12

12 488 5 %F diet survey R F R FR F (for reference man) H R F /4 69% 29:F 73% 49: 67% 175:d H H 55% 2:HH 42% 3:H 6% 1:H :%F 8:%R 9:%R99% %R %R 46% %F -59% -97% 7-33% 18% 4 larke et al., % % X ( ) H R X F X X ( )

13 489 5 %F ( ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 13.7 R H F F ean ( ) 16.5* 2.5* (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) H -31 * 12 R 38 * a * 14 Rmn F 3 5 c%h2-32 * mn F c% -3 8 c R 9 2 F d -34 * 18 HRc 31 * 1 -R R * 68 df Hc R 68 7 %R F 37 * 94 dh c -36 * 15 -H %R F 31 * 97 drhr Hc H %R F dr c -36 * 19 -R 53 1 R H dr Rc F -36 * 15 H /R dh 39 * 35 mc 42 3 e Fe %F dz Rc FRR u -37 * 88 %F d -33 * 38 RXc RF / d Rc g 32 * 91 / 34 * 164 df 3 42 c m F n df c >2m F 5 93 n dfh Rc -27 R2 RFF %F n df c -36 * R9 14: %F %n df Rc -4 R1 16: R -38 * 96 %n dr c 5 R14 24: R %n dfr RRHb -45 R15 16:1n %R : dfh RRHc 3 R21 2:5n %R %14: d Rc 28 R26 2:4n6 38 * 37 R %16: d 35 * 97 Rb l/cre HFR %18: d 36 * 98 Rc -34 * 2 /cre HR %18: e 34 * 17 a -6 3 a/cre HR -38 * 147 %18:2-31 * 247 fadj Ha R/cre %18: * 119 Ra -38 * 7 R/cre 73 5 R arr -34 * H 42 9 R/cre R dr R 55 2 HH /cre FH er 39 5 H ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. ogether, %fat (5:%F), % plant protein (8:%R) and % carbohydrate (9:%R) make up 99% of dietary calories, so increases in any of these three percentages must imply decreases in another. espite this, %R is positively correlated (46% ) with %R, so both of them are strongly negatively correlated (-59% and -97%, respectively) with %F. Higher along the coast and in the northern herding areas. ide range of values (7-33%; mean equals 18%) that are all lower than the mean of about 4% (larke et al., 1997). he wide range of values is reliable, as there is a very good correlation between xiangs (81%, and a good correlation with 1983 values (55% ). he very low values of about 1% in the three counties in Henan (province ) are reliable, as they occur in both xiangs (and were also seen in the 1983 survey). he relatively high values of about 3% in some of the counties in uangdong (province ) are reliable as they were seen in both xiangs but represent a real increase since ncrease by one quarter in overall mean between 1983 and 1989 surveys probably represents a real increase in percentage fat in the diet. ositively correlated with variables related to animal food intake (e.g., 69% 29:F; 73% 49:; 67% 175:d) and with H, nonh and total cholesterol (55% 2:HH; 42% 3:H; 6% 1:H). his variable correlates with so many factors that causal relationships are difficult to assess. R : : 11-12

14 49 7 %R diet survey R F R FR R (for reference man) H R F X ( ) H X R F X X ( )

15 491 7 %R ( ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F. R H F F ean ( ) (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) * 11 Z-R 5 2 F FH d R R F -37 * 68 df 4 25 c H 68 5 %F * 93 dr Rc * 6 %R * 94 dh mc R %R F dfadj 33 * 36 fc HF %R F 39 * 117 drrh 35 * 37 Rc -31 * 23 H 69 1 R H drday 38 * 39 Rc 55 3 e -33 * 15 H /R dday 28 5 b 36 * 33 FRR RF %F dr Rc RF -37 * 19 Fe %F -33 * 158 dh b 35 * u 37 * 89 %F -37 * 159 dz -32 * 57 c m -35 * / d -36 * 58 b >2m g 37 * 91 / dfh RHHb -33 * R2 RFF err n df RHHc -26 R9 14: F %n dr -38 * 64 Rb -38 * R1 16: F %n dfr Rb 71 R14 24: %F 36 * 97 %n dfh Rc -25 R15 16:1n %F : d Rc 66 R21 2:5n R %14: d c -27 R25 2:3n R %16: d RRHb 28 R26 2:4n %R %18: 5 21 e 3 81 Rc l/cre %R 34 * 146 %18:1-33 * 247 fadj 37 * 82 c -34 * 2 /cre HFR %18:2-32 * H a/cre -38 * 39 HR -34 * 148 %18: HH -4 7 R/cre H -31 * 17 arr R 55 3 H 43 8 R * 19 dr 36 * 5 H R/cre a mn 76 5 R 37 * 52 c% R -34 * 1 35 * 51 R d ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. xcellent correlation between xiangs (91% ) and with 1983 (88% ) implies real geographic differences. ome increase in mean value since 1983, but even the highest values in 1989 are still well below western average. orrelated positively with variables related to animal food intake (e.g., 88% 29:F; 78% 49:; 7% 175:d) and with H, non-h and total cholesterol (57% 2:HH; 55% 3:H; 71% 1:H) % % % 29:F 78% 49: 7% 175:d H H 57% 2:HH; 55% 3:H; 71% 1:H R : : 11-12

16 492 8 %R diet survey R F R FR R (for reference man) H R F % % 5:%F 8:%R 9:%R99% %R %R 46% %F -59% -97% 62% 38HFR 66% 158:dH -61% 5:R -54% 174:dFH -41% 4:1 X ( ) R HX F X X ( )

17 493 8 %R ( ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 9.7 R H F F ean ( ) 9.1* 8.1* (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) Fc H mn * 149 dr c 34 * * 26 a drday 31 * 18 HRc F F -33 * 153 dday -35 * 19 RHb R 5 3 R F -36 * 156 dday c R %F H dr Rb HF 62 6 %R %F dh -33 * 31 Rc 38 * 23 H %R %F dz 34 * 38 RXc F 46 9 %R %F 37 * 161 d Hc e -4 1 R 43 9 / 36 * 162 d 3 58 b 3 32 Fe H / df 34 * 59 c RF 55 2 u n dfh 34 * 63 Hc m %n df 39 * 64 Rb Hsg g %n dr 3 65 Rc 25 R3 F F %n9-31 * 173 dfr 34 * 67 -Rc 4 R9 14: %F : dfh 32 * 77 c 36 * R1 16: %F %16: d -31 * 81 Rc 31 R11 18: R %18: d -38 * 82 c -45 R14 24: R %18: ehr -36 * 89 c 3 R15 16:1n %R %18: e Rb -3 R21 2:5n %R %18: e Rc 26 R23 18:2n R arr e%r a -25 R26 2:4n HFR -32 * 31 a er 35 * 111 a 52 1 l/cre HR c% 39 * 243 fadj Ha 41 2 /cre R 5 57 d fadj Ra 53 3 a/cre FR d H 65 6 R/cre dhh * 2 HH 35 * 7 R/cre R dh 58 4 R R -34 * 51 R d H R/cre FH -35 * 93 dr -36 * 7 R /cre -32 * 53 F dfadj R R F drrh Z-R Rmn F dm R : : ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. ery strong correlation between xiangs (91% ) and good correlation with 1983 values (62% ). ogether, %fat (5:%F), % plant protein (8:%R) and % carbohydrate (9:%R) make up 99% of dietary calories, so increases in any of these three percentages must imply decreases in another. espite this, %R is positively correlated (46% ) with %R, so both of them are strongly negatively correlated (-59% and -97%, respectively) with %F. Higher in the north, hence correlated positively with variables related to wheat and other plant food intake (e.g., 62% 38HFR; 66% 158:dH) and negatively with variables related to rice and animal food intake (e.g., -61% 5:R; -54% 174:dFH; -41% 4:1).

18 494 9 %R diet survey R F R FR RHR (for reference man) F R H % % 5:%F 8:%R 9:%R99% %R %R 46% %F -59% -97% 5:%F -97% 45% 28:F 38%* 158:dH -72% 5:R -61% 174:dFH -6% 4: X ( ) F X X R H X ( )

19 495 9 %R ( ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 76.6 R H F F ean ( ) 73.2* 68.9* (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) H R FH c% Rmn -33 * * 68 df 32 * F -39 * 94 dh 31 8 c -37 * 9 -R F 5 59 F dh b R 67 4 R 35 * * 97 drhr 33 * 18 HRc R %F drrh Hc 32 * 15 -H %R F dr c %R F drday Hc 34 * 19 -R R F dr c e H H 38 * 158 dh -34 * 31 Rc -34 * 33 FRR 36 * 19 Fe /R 33 * 159 dz Rc RF 43 2 u -37 * 87 %F d mc * * 88 %F df fc g 28 9 / df Rc -37 * 41 m Zn / dfh RXc 33 * 47 >2m F n df -37 * 39 Rc 31 * R2 RFF F n dr c 31 * R9 14: %F n dfr b 38 * R1 16: %F 36 * 96 %n dfh c -53 R14 24: R -33 * 97 %n d 3 65 Rc 4 R15 16:1n R : d c -38 * R21 2:5n %R %14: d 32 * 75 Rc 25 R25 2:3n %R %16: e c -25 R26 2:4n R -36 * 145 %18: fadj RRHb 62 1 l/cre 34 * 38 HFR -32 * 146 %18: * 79 RRHc 31 2 /cre 36 * 39 HR 36 * 147 %18: Rc 58 3 a/cre 25 4 HR %18:3 33 * 4 R -36 * 98 Rc 33 * 6 R/cre arr -32 * 5 H b 38 * 7 R/cre dr -32 * 17 a R er H R/cre R a HH /cre R -33 * 5 c%h2 R : : ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. ery strong correlation between xiangs (81% ) and with 1983 values (64% ). ogether, %fat (5:%F), % plant protein (8:%R) and % carbohydrate (9:%R) make up 99% of dietary calories, so increases in any of these three percentages must imply decreases in another. espite this, %R is positively correlated (46% ) with %R, so both of them are strongly negatively correlated (-59% and -97%, respectively) with %F. Higher in the north. uch higher percentage intake from carbohydrates than in western diets. his variable is almost perfectly negatively correlated (-97% ) with the percentage of calories derived from fat (5:%F), so its correlates are almost the same (with the reverse sign) as with that variable. orrelated positively with variables related to wheat and other plant food intake (e.g., 45% 28:F; 38%* 158:dH) and negatively with variables related to rice and animal food intake (e.g., -72% 5:R; -61% 174:dFH; -6% 4:1).

20 F H R X X X ( ) X ( ) F H R R diet survey R (retinol equivalents/day/reference man) 496

21 R ( / / ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 339 R H F F ean ( ) (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) Hb mn -35 * dr b R a -32 * 74 H dh 4 19 RHb a -3 6 %R HR d 28 2 RHc RFa /R dr Hc H : dfr -38 * 41 b -36 * 3 H 32 * %14: d Rc R a -36 * 17 arr d b Z-R a 33 * 52 c% dr Rb -32 * * 25 a -33 * 9 dhh -33 * 184 d Rb RR -32 * 26 err -32 * 91 dh d Rc HF F d ehrr Rb F 3 37 R 33 * 96 dr er b e -32 * 38 HFR dfadj c -34 * 33 FRR 46 4 HR dr -3 4 R -41 * 95 Rb -25 R1 Hb R drday Rc -3 R13 22: R 31 * 156 dday ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. o simple geographic pattern, but tenfold geographic variation that is validated by the close correlation (86% ) between xiangs and the moderate correlation (47% ) with 1983 values. espite reasonably reliable evidence as to which counties have had persistently high or low carotenoid intake, no strong correlations with cause-specific (or overall) mortality are apparent. s carotenoids make up 9% of total vitamin intake (expressed in retinol equivalents) in rural hina, there is a close correlation (99% ) with 12:, so the pattern of geographic variation of these two variables is very similar. orrelated with plant food intake, including sweet potatos, a major source of carotenoids (e.g., 43% 28:F; 75% 14:; 53% 163:d) % % 9 12: 99% 43% 28:F 75% 14: 53% 163:d R : : 11-12

22 diet survey (retinol equivalents/day/reference man) H F R X ( ) F H X X R X ( )

23 ( / / ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 339 R H F F ean ( ) (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) b Ra a R 33 * 157 dr HRc -31 * 3 H %R -36 * dh RHb -34 * 11 Z-R R H d 29 2 RHc H arr dr Hc -32 * RF c% dr -36 * 41 b RR 34 * d d Rc HF a -34 * 9 dhh d Hc H 31 * 24 a dh ehrr -37 * 64 Rb F 36 * 25 a d er Rc FRR err 32 * 96 dr c -27 R1 Hb F dfadj * 95 Rb -31 * R13 22: 33 * 37 R drrh -32 * 4 R Rc -24 R22 22:6n3-33 * 38 HFR 31 * 149 dr Hb F1/cre 43 4 HR 32 * 155 drday R 32 * 23 3mn H 35 * 156 dday ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. ery strong correlation between xiangs (85% ) and strong correlation with 1983 values (44% ). his variable represents total vitamin intake, from plant sources (9% of total) (see 11:R) and animal sources (1% of total), expressed as retinol equivalents (R). ne R is equal to 1 g of retinol or 6 gs of beta-carotene. ore than tenfold variation about the mean of 7, from less than 2 (in seven counties) to about 2. he reality of this wide variation is demonstration by the close correlation (85% ) between xiangs. here is also a moderate (41%*) correlation with 1983 values. lthough the overall mean consumption in rural hina is below consumption, and the lowest county values (below 2) might, if they persisted, be expected to lead to vitamin deficiency, there is no significant geographic correlation with the measured plasma levels of retinol (or of beta-carotene, the main provitamin in this population), and no convincingly negative geographic correlations with cause-specific mortality (in fact, the closest such correlation [-39%*] is for death from road traffic accidents at ages -34 [95:Rb] % % 9% 11:R 1% R 1R 1 g 6 g % %* 2 34 [-39%*] [95:Rb] R : : 11-12

24 F H R X X X ( ) X ( ) F H R diet survey (mg/day/reference man) 5

25 51 13 ( / / ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 19.6 R H F F ean ( ) (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) c 47 R18 22:1n %F dh -36 * * 73 b 34 * R19 24:1n / 4 97 drhr -32 * * 74 c -3 R22 22:6n / d b -34 * 75 Rc -32 * R25 2:3n n3 33 * 112 dfadj -33 * 11 Fb RRHb 25 8 R n df -35 * 12 Fc RRHc %n df c Rb 39 2 F %n df Hc Rmc 43 3 R %n d c Rc RF %14: d c Rc 31 * %16: drh c a 33 * 2 u %18: e 39 * 29 Rc -33 * 16 a n %18: erh Rc RFa F %18: e3rd 39 * 33 c a 33 * 33 R %18: e mc -35 * 117 a * 7 chhz ef Rc F arr 35 * 243 fadj 36 * 4 Hc -3 7 R er 36 * 245 fhadj 35 * 42 c F 38 * 31 a 33 * Rb R 3 78 HR 33 * 52 c% -36 * 5 H c -39 R1 Hb 4 79 RH -33 * 69 d b -25 R6 n F 28 9 dhh Rb -28 R15 16:1n %F dh b -44 R16 18:1n %F d ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. trong geographic clustering around hanghai. enfold range of county means, validated by strong correlation (77% ) between xiangs: the overall mean is 25 mg/day, but the county means range from about 5-5 mg/day. ven the lower levels, however, may well suffice to avoid deficiency. trong correlation with vegetable oil intake (91% 54:) o significant correlation with plasma vitamin levels, so any associations with disease may be non-causal. oderately strong positive association with indices of prosperity (e.g., 38%*, 31:a) and hence moderately strong negative correlations with diseases of poverty in adult life (e.g., and other infections; -45%, 16:c, -35%*, 12:Fc) and with childhood mortality, but not with vascular or neoplastic diseases % 25 mg/ 5-5 mg/ 91% 54: 38%* 31:a -45%, 16:c, -35%*, 12:Fc R : : 11-12

26 F H R X X X ( ) X ( ) F H R diet survey (R ) (mg/day/reference man) 52

27 53 14 ( ) ( / / ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 74.4 R H F F ean ( ) (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) 3 1 c R d 33 * 12 Fc 38 * 21 RR dfadj b 31 * 24 F F drrh 3 16 c 32 * a * 149 dr 32 * 19 RHb e a -32 * 86 /R 34 * 155 drday 34 * 47 Rc -35 * a : 34 * 156 dday b 31 * 42 Hsg err %14: dr 36 * 69 Rc -31 R3 F F dr 36 * 163 d 34 * 72 c 29 R4 F F c%h dr c -35 * R13 22: %F c%fh d Rb 29 R19 24:1n %F -34 * 9 dhh d b -33 * 6 R/cre 34 * 37 R dh d 33 1 c 31 * HFR dr ehrr H 34 * 4 R 54 4 HR 35 * 96 dr H %R R -32 * 18 d ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. ide range of county means, confirmed by strong (75% ) correlation beween xiangs, but even the lowest county intakes do not suggest vitamin deficiency. he recommended daily allowance (R) for vitamin (for an adult male) is 9 mg/day, and the estimated average requirement (R) is 6 mg/day. o clearly significantly negative geographic correlations between vitamin intake levels and cause-specific mortality % R 9 mg/ R 6 mg/ R : : 11-12

28 F H R X X X ( ) X ( ) F H R H diet survey H ( 1) (mg/day/reference man) 54

29 55 15 H ( 1) ( / / ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 2.1 R H F F ean ( ) (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) * 6-33 * 7 %R F -37 * 93 dr b R 71 8 %R dh -32 * 12 Fc %R d c R R F dfadj -33 * 22 b R RF -38 * 84 F drrh c F H dm Hc m 59 2 u %F drday Rb Hsg %F -35 * 153 dday -32 * 31 Rc 27 R2 RFF n %F -35 * 156 dday Hc 24 R9 14: Zn 35 * 9 / dr b 25 R1 16: F / dh c -27 R14 24: 3 31 %F n dz -3 8 Rb -31 * R16 18:1n %F 32 * 95 %n d -37 * 81 Rc 51 1 l/cre R %n6-39 * 164 df -36 * 82 c 31 2 /cre -34 * 34 R %n9-35 * 166 dfh c 5 3 a/cre %R %16:1-31 * 167 df Rb 67 6 R/cre %R %18: -32 * 168 df Rc 29 7 R/cre R %18: dr -34 * 99 b /cre HFR %18: dfh -37 * 1 c 32 * 12 R HR 36 * 148 %18: d Hc 33 * 14 Rmn FR chhz drh F -31 * 23 3mn dr fadj a -33 * 2 F -38 * 5 R 38 * 51 c%fh fadj a 63 3 R -32 * 51 R d Ha 26 4 R FH 37 * 9 dhh Ra -36 * 5 %F -36 * 53 F dh 45 4 R -3 2 HH 48 6 %R -32 * 55 F 36 * 92 d H ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. trong (and persistent) fourfold variation, confirmed by strong correlation between xiangs (88% ) and with 1983 values (66% ). s the geographic variation is dominated by north-south variation, this variable has many other correlates that may be only indirectly associated % % R : : 11-12

30 ( ) F H R X X X ( ) X ( ) F H R (uoli) not plotted (1983): 2.7 (1989):.7 16 RF diet survey RF ( 2) (mg/day/reference man) 56

31 57 16 RF ( 2) ( / / ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F.7 R H F F ean ( ).7.7 (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs * ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) * 78 RRHb 25 7 %R R a -33 * 16 c -32 * 8 Rb 38 * 1 R 3 34 R 33 * 52 c% 26 2 RHc Rc R -32 * 69 d 37 * 21 Hc a * dh 36 * 29 Rc a H FH d Rb Ra dh Rc a 38 * dh 33 * 35 mc R 32 * 2 u 38 * 78 HR drhr Rc R RH d Rb Hsb 4 23 n 35 * 83 F d Rb 28 R18 22:1n a H 4 21 e 25 5 b 31 R21 2:5n err 33 * 86 /R ehrr -34 * 73 b Zn n e%fr c 27 2 F 34 * 29 F 34 * 93 n Rc 56 3 R %F : c 33 * 6 %R %F arr ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. trong geographic clustering around hanghai R : : 11-12

32 58 17 diet survey (mg/day/reference man) R H F X ( ) H X F X R X ( )

33 59 17 ( / / ) nland rovinces ( ) oastal rovinces ( ) rea ntake rea ntake rea ntake rea ntake rea ntake X X F 12.4 R H F F ean ( ) (a) (b) ean (a) (a) ean (b) (b) r% t-test Xiang ( ) vs Xiang ( ) vs ll 2 <.5 correlations (r%) with other items ( ): * 2 <.1, 2 <.1 Full variable names are in ummary tatistics, pp19-13 ( ) Rc 3 R4 F * arr 34 * 32 Rc -27 R7 F H dm 33 * 4 Hc 25 R17 2:1n RF RH d<25m b 32 * R18 22:1n a 36 * 83 F 33 * 149 dr RHHc 32 * R19 24:1n u %F drday Rc -26 R26 2:4n / dday Rc /cre 31 * 23 n / dday Ha a n df Ra 29 3 R 33 * 25 a 36 * 93 n6 34 * 169 df R 32 * 11 R 31 * 28 F -32 * 14 %16: d F 34 * HR -34 * 7 chhz m 31 * * arf ietary intakes are standardised per 'reference man', defined as an adult male, years of age, 65 kg of body weight and undertaking very light physical work. he 1983 diet survey in uoli (ounty ) took place during a festival when an atypical diet was consumed, so is excluded from the 1983 vs 1989 correlations. he geographic variation is real (73% correlation between xiangs), but is dominated by the high values in three neighbouring counties in iangsu (province ) % 3 R : : 11-12