探討體位變化對死亡率的影響-北台灣世代研究資料

Impact of Weight Status Change on Mortality: A Prospective Cohort Study in Northern Taiwan

盧佳文(Chia-Wen Lu)

指導教授 ： 陳秀熙

共同指導教授 ： 黃國晉

國立臺灣大學/公共衛生學院/流行病學與預防醫學研究所/碩士(2013年)

https://doi.org/10.6342/NTU.2013.01584

摘要

背景
雖然已有許多研究探討肥胖與死亡率的相關性，但缺乏長期追蹤資料下年齡-性別特定的，或考慮其它共變數的體位狀態（依身體質量指數(BMI)分類）的盛行率，及其隨時間的變化，還有在這種動態變化下，量化與預測不同體位對死亡率的相關性。
目的
本研究目的為利用長期追蹤資料去探討年齡-性別相關的體位狀態(依BMI分類)的盛行率，及其隨時間的變化，並預測這種動態變化對全死因及肥胖相關死因的影響。
材料與方法
本究資料來源得自北台灣整合式篩檢計畫，研究對象為1999年至2009年20歲以上參與此計畫的民眾。在第一次篩檢時，利用問卷調查瞭解參與者的基本資料、生活習慣與個人病史。在第一次篩檢與之後每一次再篩檢時，量測身高、體重、血壓、個人習慣變化、並抽血。體位依衛生署的標準，其依身體質量指數(BMI)將體位分為體重過輕、體重正常、體重過重、與肥胖四類。利用非時間相依與時間相依兩種Cox風險比例迴歸模型去評估基準體位對全死因及肥胖相關死因的風險比，並估計在調整其它如年齡、性別、教育程度等干擾因子後的風險比。最後，利用多階段馬可夫鏈，在考慮體位具動態變化的情況，預測不同體位對全死因及肥胖相關死因的死亡率。
結果
共有108,616人參與篩檢，其中的60.2%為女性，平均年齡為48.2± 15.4歲。其中約55%的人體重為過重或肥胖，肥胖者約占1/4，過輕者占4%。約四成民眾的體重在觀察期間沒有改變。肥胖年齡別盛行率呈倒U型，從20-20歲的20%開始上升到50-59歲時的37%，然後到80歲後下降到28%。體重過重年齡別盛行率亦有相似的趨勢，其自20-20歲的23%開始上升到50-59歲時的26%，至80歲後下降到18%。體重過輕年齡別盛行率則呈Ｕ型，由從20-20歲的6%開始下降到50-59歲時的1%，然後到80歲後又回到6%。值得注意的是20-29歲女性中，其體重過輕者占絕大多數。高教育程度與體重過重及肥胖者具逆相關，但與體重過輕具正相關。馬可夫模式可允許不同的時間間隔下的測量值去估計狀態間的轉移率，依此方法所估計到由正常到體重過重的年轉移速率為9%;由體重過重到肥胖的年轉移速率為10%；由正常到過輕的年轉移速率為2%。而由過輕到正常、由過重到正常、由肥胖到過重的年轉移速率分別為24%、12%、12%。
在考慮年齡、性別、教育程度、抽菸、運動、高血壓、糖尿病等干擾因子的效應後，存活率最差者為體重過輕組，然後是正常體重組、肥胖組；最好的是體重過重組。校正依時間而變的體位變化後，時間相依Cox迴歸模式估計結果顯示：以體重正常組為基準，體重過輕組在全死因死亡的相對危險性為1.85 (95% CI:1.63-2.10)，體重過重組的相對危險性為0.80 (95% CI:0.75-0.86)，肥胖組為0.84 (95% CI:0.79-0.91)。依多階段馬可夫，將體位變化納入模式分析的結果亦一致，同樣以體重正常為基準時，體重過輕組對全死因死亡的相對危險性約為2倍 (adjusted HR=2.01 (95% CI:1.72-2.35))；過重組和肥胖組則分別降低39%(adjusted HR=0.61 (95% CI:0.54-0.69)和14% ( adjusted HR=0.86 (95% CI:0.79-0.93) 的死亡風險。
結論
利用BMI進行體位分類，分成過輕、正常、過重、肥胖四組時，約40%的人會維持在正常體重。肥胖者占1/4，過輕者占4%。肥胖與過輕的盛行率與年齡的關係非呈線性而是Ｕ型。20-20歲的年輕女性的體重過輕者占大部分。進展速率和恢復速率顯現出體位的動態變化與體位不良的嚴重度呈正比，其亦可用來預測死亡風險。在調整年齡、性別、教育程度的干擾作用下，體重過輕者的死亡風險約為正常者的二倍；過重及肥胖者的風險則比正常體重者低。由本研究結果可以提供台灣民眾的依性別-年齡別而異的適當體重分類切點的建議。而體位的動態變化與死亡之間的關係亦可作為評估某項介入計畫欲達到正常體重時，其在基準點時的資訊。

關鍵字

身體質量指數 ； 體位 ； 肥胖 ； 死亡率 ； 存活分析 ； 馬可夫模型

並列摘要

Background
Although a series of studies reporting prevalence of obesity and the relationship between obesity and mortality, there is lacking of a longitudinal follow-up study to investigate age-gender-dependent or other covariate-based prevalence of weight status defined by body mass index (BMI) categories, weight status changes with time, and also to quantify the effects of dynamic weight status to mortality or the prediction of mortality by weight status changes.
Aims
To investigate age-gender-dependent prevalence of weight status, estimate the change of weight status with time, and to examine the relationship between the dynamic weight-changing statuses using BMI and all-cause of deaths and obesity-related deaths based on a longitudinal follow-up study.
Methods
Data used for the analysis were derived from an integrated model of community-based multiple screening in northern city of Taiwan. Subjects aged 20 years or older between 1999 and 2010 were enrolled. By annual visiting and staggered entry, a baseline questionnaire was administrated to obtain the demographic characteristics, lifestyle variables and personal history in first visiting. In the first and following visiting, height, weight, blood pressure, life style changes and blood sampling were collected and performed by standard devise and trained staff. The categories of BMI were divided into underweight, normal weight, overweight, and obesity groups based on the criteria set by the Department of Health, Executive Yuan, Taiwan. Time-invariant and time-dependent Cox hazards regression model were used to assess the impacts of baseline weight status and also dynamic change of weight status on hazard rate for the risk for death with adjustment for age, gender, education level, and other possible confounding factors. Markov model was further applied to estimating and predicting the all-cause death and obesity-related death in association with dynamic change of weight status.

Results
Among the invited subjects, a total of 108,616 cases were enrolled into screening. Of these subjects, 60.2% were female; the mean age was 48.2± 15.4 years. The overall community residents had approximately 55% with BMI above normal (overweight plus obesity), around a quarter of obese residents (25%), and 4% underweight. Around 40% of the community residents have kept the weight within normal range between 1999 and 2009. Age-specific prevalence of obesity showed an inverse U-shape profile increasing from 20% in 20-29 years of age group to 26% in 50-59 years of age group, and then decreasing to 18% in those aged 80 years or older. The similar pattern was noted for age-specific overweight from 23% in 20-29 years of age group to 37% in 50-59 years of age group, and then decreasing to 28% in those aged 80 years or older. The age-specific prevalence of underweight showed a U-shape profile decreasing from 6% in 20-29 years of age group to 1% in 50-59 years of age group and then rising to 6% in those aged 80 years or older. The similar pattern was noted for females but underweight predominated in the young adults aged 20-29 years. High education level was inversely associated with overweight and obesity and but positively associated with underweight. With the advent of multistate Markov model allowing for different intervals between the two visits annual progression rates were 9% for the transition from normal weight to overweight and 10% from overweight to obesity. Annual rate for the transition from normal weight to underweight was only 2%. The corresponding regression rates were 24% for underweight to normal, 12% for overweight to normal weight and also obesity to overweight.
Regarding the effect of weight status on mortality, after adjusting age, gender, education level, smoking, history of hypertension and diabetes, the worst survival was noted in the underweight group, followed by the normal weight group, the obesity, and the lowest in the overweight group. Considering the time-varying changes of weight status, the underweight group conferred 1.85-fold (95% CI: 1.63-2.10) risk for all-cause death whereas the overweight group (adjusted HR=0.80 (95% CI: 0.75-0.86)) and the obesity group (adjusted HR=0.84 (95% CI: 0.79-0.91)) had a lower risk comparing to normal weight group based on time-varying Cox regression model. Taking dynamic changes of four weight statuses into account, the results of multi-state Markov models show the consistent findings that underweight led to two times (adjusted HR=2.01 (95% CI: 1.72-2.35)) likely to die compared with the normal weight and overweight and obesity has a lower risk for death by 39% (adjusted HR=0.61 (95% CI: 0.54-0.69)) and 14% (adjusted HR=0.86 (95% CI: 0.79-0.93)) compared with normal weight.

Conclusions
Using the four categories of BMI to define weight status (underweight, normal weight, overweight, and obesity), we found that 40% of the community residents were within normal range and around a quarter of residents were obese and 4% were underweight. The relationship between the prevalence of obesity and underweight versus age was non-linear (U-shape). Underweight preponderated in young or female adults aged 20-29 years. Progression and regression rates show such dynamic changes were proportional to the severity of weight status and can be predicted. Underweight led to two times likely to die compared with the normal weight; and overweight and obesity has a lower risk for death compared with normal weight after taking age, gender, and education level into account. These findings of the effect of weight status on mortality would be useful for determining the optimal age-gender-dependent cutoff for re-defining overweight, obesity, and underweight for Taiwanese people. The dynamic change of weight status in relation to mortality also provides baseline information when certain specific intervention on the attainment of normal weight is assessed.

並列關鍵字

BMI ； weight status ； obesity ； mortality, ； survival analysis ； Markov model

參考文獻

1. Wang, Y.C., et al., Health and economic burden of the projected obesity trends in the USA and the UK. Lancet, 2011. 378(9793): p. 815-25.

2. Whitlock, G., et al., Body-mass index and cause-specific mortality in 900 000 adults: collaborative analyses of 57 prospective studies. Lancet, 2009. 373(9669): p. 1083-96.

3. Greenberg, J.A., The obesity paradox in the US population. Am J Clin Nutr, 2013. 97(6): p. 1195-200.

4. Lobstein, T., L. Baur, and R. Uauy, Obesity in children and young people: a crisis in public health. Obes Rev, 2004. 5 Suppl 1: p. 4-104.

5. Pan, W.H., et al., Body mass index and obesity-related metabolic disorders in Taiwanese and US whites and blacks: implications for definitions of overweight and obesity for Asians. Am J Clin Nutr, 2004. 79(1): p. 31-9.