南京居民膳食模式与高血糖关系的随访研究

Dietary patterns and hyperglycemia in a follow-up study in Nanjing city

目的探索南京居民不同膳食模式与高血糖发病风险的关系。方法采用多阶段分层随机整群抽样方法，于2007年6-9月对南京市2个城区的7个社区内的30岁以上的常住居民开展基线调查，符合条件的调查对象共3376名，剔除已确诊的高血糖患者（476例），将2900例非高血糖的人群作为本研究的基线样本，于2010年6—9月对非高血糖人群进行随访调查。采用专门设计的食物频数量表（FFQ），应用因子分析方法建立膳食模式，采用多因素线性和Cox回归模型分析不同膳食模式与高血糖发病风险的关联。结果3年后共随访2093名，随访率为72．2％。高血糖3年累计发病率为7．5％（158／2093）；男女发病率分别为7．1％（62／873）、7．9％（96／1220），差异无统计学意义（x^2=0．43，P=0．513）。通过因子分析得到5种膳食模式：调味品膳食模式、动植物蛋白膳食模式、传统健康膳食模式、甜食膳食模式和饮酒膳食模式。经多元线性回归分析，传统健康膳食模式、甜食膳食模式因子得分平均每增加1个单位，3年后随访FBG值分别增加-0．054、0．050mmol／L，差异均有统计学意义（t值分别为-2．38、2．27，P值均〈0．05）。经多因素Cox回归分析，男性甜食膳食模式与高血糖发病风险呈正关联；最低三分位（T1）和最高三分位（T3）的高血糖发病率分别为4．7％（14／295）、9．7％（26／269）（T3与T1相比，RR=1．88，95％CI：1．04～3．54）；女性传统健康膳食模式与高血糖发病风险呈负关联，T3、T1发病率分别为6．3％（21／335）、10．7％（45／421）（T3与T1相比，RR：0．59，95％CI：0．35—0．99），而饮酒膳食模式与高血糖发病风险呈正关联，T3、T1发病率分别为11．1％（33／297）、8．1％（38／472）（T3与T1相比，RR=1．35，95％CI：0．84—2．16）。结论膳食模式与高血糖密切相关。甜食膳食模式是男性高血糖发病的危险因素；传统健康膳食模式是女性的保护因素，而饮酒膳食模式是女性的危险因素。

Objective To investigate the association between dietary patterns and the risk of developing hyperglycemia in Nanjing. Methods Using multi-stage stratified random cluster sampling, the baseline survey was conducted on local residents older than 30 years in 7 communities from 2 urban districts from June to September 2007 in Nanjing. The total eligible subjects were 3376. Excluding the 476 previously diagnosed hyperglycemia patients ,2900 non-hyperglycemia subjects were used as the baseline sample for the follow-up survey from June to September 2010. Using specially designed food frequency questionnaire （FFQ） ,factor analysis was applied to identify food patterns. Muhivariable linear and Cox regression models were used to analyze the association between different dietary patterns and risk of hyperglycemia. Results The follow-up rate was 72. 2% , with 2093 subjects participated the follow-up survey in 3 years. Three-year cumulative incidence of hyperglycemia was 7.5% （ 158/2093 ）. The incidence rate was 7.1% （ 62/873 ） for males and 7.9% （96/1220） for females, but the differences were not statistically significant （ X^2 = 0. 43, P = O. 513）. Five dietary patterns were identified by factor analysis： condiment, animal and plant protein, traditional healthy, sweet food and alcohol drinking. By multivariable linear regression, on average, an increase in traditional healthy pattern and sweet food pattern of 1 unit was associated with a -0. 054,0. 050 mmol/L increase in fasting blood glucose, respectively, and the differences were both statistically significant （t = -2. 38,2. 27, respectively,P values were both less than O. 05）. By multivariable Cox regression, the pattern sweet food was positively significantly associated with hyperglycemia risk in men. The incidence of hyperglycemia was 4. 7% （ 14/295 ） for the lowest tertile of the factor score （ T1 ）, and 9.7% （ 26/269 ） for the highest tertile of the factor score （T3） （T3： TI： RR = 1.88,95% C1：1.04 -3.54）.The pattern traditional healthy was inversely associated with hyperglycemia risk in women. The incidence of hyperglycemia was 10. 7% （45/421） for T1 and 6.3% （21/335） for T3 （T3：T1 ： RR =0. 59,95% CI： 0. 35 -0. 99）. Conversely, a statistically significant positively association was observed for the pattern alcohol drinking in women. The incidence of hyperglycemia was 8. 1% （ 38/472 ） for TI and 11. 1% （33/297） for T3 （ T3 ： T1 ： RR = 1.35,95% CI： O. 84 - 2. 16 ） . Conclusion Dietary patterns are associated with hyperglycemia. The sweet food pattern is a risk factor for hyperglycemia in men. In women, healthy dietary pattern is healthy and the alcohol drinking pattern is a risk factor for hyperglycemia.