臺灣博碩士論文加值系統

前言:利用除去口腔癌前病兆，如口腔白斑，來預防口腔癌，是一種可行的防治策略。但
必須了解從正常到白斑及口腔癌的自然病史，方能評估各種危險因子對發生白斑及 其轉
變成口腔癌的影響。而先前研究常因個案數太少、追蹤時間過長、及受到白斑治 療的影
響，而有所限制。本研究提出另一種回溯性追蹤方式，來為克服上述問題。 目標：本
研究之目的為：(1)估計危險因子對白斑發生的影響，(2)推估危險因子對促 使白斑轉變
成癌症的影響，(3)利用三階段馬可夫鏈模式，同步估計白斑發生及轉變成癌症的速率。
材料與方法：從台
北某一家醫學中心，回溯追蹤自1988至1998間三個疾病世代，利用 口腔白斑及牙周病兩
世代，以配對病例對照研究方法，探討第一個研究目的，以白斑 世代資料連結癌症登記
資料及病歷回顧方式，以探討第二個研究目的。發展三階段病 例對照研究方法，由口腔
白斑、口腔癌及牙周病三世代中抽樣，探討第三個研究目的 。統計方法為：利用條件回
歸分析方法(conditional logistic regression)估計危險因子對白斑發生的影響，使用
兩個存活分析模式估計口腔白斑轉變癌症速率及危險因 子的影響程度，及發展三階段馬
可夫鏈模式，同時估計口腔白斑發生率及其轉變癌症 的速率。
結果：經分析資料結果顯示：嚼食檳榔
及吸菸對口腔白斑發生之危險比分別為34(95% CI:4.5~259.0)及3.6(95%CI:1.0~12.5)。
使用偉伯存活分析模式，發現白斑轉變癌症 速率會隨追蹤時間增加而上升。經年齡及性
別調整後，僅有嚼食檳榔會加速轉變速率 ，其相對危險速率比為5.04(95%CI:1.27~20.06
)。利用三階段病例對照方法，同時估 計白斑發生率為0.18%(95%CI:0.15%~0.21%)及其轉
變癌症速率為11.1%(95%CI:9.27%~ 13%)。
結論：從實務面看，本研究發現嚼食檳榔不僅會引起白斑
的發生且會加速其轉變成癌症。若除去嚼食檳榔習慣，可減少人群中77%白斑病人的發生
。另白斑的治療亦可下降76%的白斑轉變成癌症。這顯示若每年口腔檢查白斑一次，可降
低57%口腔癌的發生。從方法學看，三階段馬可夫鏈模式應用於回溯性世代之病例對照方
法對疾病轉變速率的估計，將可運用於其他慢性疾病研究。

Introduction: To prevent oral cancer, elimination of precancerous lesion i.e.l
eukoplakia is feasible preventive strategy. It is necessary to make a clearund
erstanding of natural history from normal, leukoplakia, to oral cancer toknow
the efficacy of dental check-up for leukoplakia in reducing oral cancer.But, s
everal limitations in previous studies are rare cases, long-term follow-up, an
d treated effect of leukoplakia. An alternative approach is proposed totackle
these problems.Objectives: The aims of this thesis are (1) to estimate the eff
ect of relevantrisk factors on the occurrence of leukoplakia; (2) to investiga
te how riskfactors affect malignant transformation ; and (3) to estimate the i
ncidenceof leukoplakia and malignant transformation simultaneously based on a
three-state Markov chain model.Material and Methods: Three retrospective follo
w-up cohorts from 1988 to 1998are identified from one medical center in Taipei
. Using leukoplakia andperiodontal disease cohort, a matched case-control stud
y is design to answerthe first objective. The leukoplakia cohort is linked to
cancer registry andchart review to identify oral carcinoma to answer the secon
d objective. Athree-state case-control study design is performed for the third
objective.A conditional logistic regression is performed to estimate the infl
uence ofrisk factors on leukoplakia. Two accelerated failure models are employ
ed toestimate the incidence of malignant transformation. A three-state Markovc
hain model adapted to case-control study is developed to estimate theincidence
of leukoplakia and malignant transformation.Results: the adjusted odds ratios
for chewing betel nut and smoking on theoccurrence of leukoplakia are 34.(95%
CI: 4.5~259.0) and 3.6 (95%CI: 1.0~12.5). Using Weibull model, the incidence
of malignant transformation ofleukoplakia increases with year. The adjusted ha
zard ratio for chewing betelnut is 5.04 (95% CI: 1.27~20.06). Using of three-s
tate case-control method,the incidence of leukoplakia for this cohort is 0.18%
(95%CI: 0.15%~0.21%),and malignant transformation is 11.1% (95%CI: 9.27%~13%)
.Conclusion: From practical viewpoint, this thesis demonstrates betel nut asa
risk factor not only for the occurrence of leukoplakia but for malignanttransf
ormation. For the occurrence of leukoplakia, getting rid of chewingbetel nut w
ould avoid 77% leukoplakia. The efficacy of treatment onleukoplakia could redu
ce 76% malignant transformation. This implies annualdental check-up for leukop
lakia could lead to 57% reduction of oral cancers.From methodological viewpoin
t, three-sate Markov chain model adapted toestimate transition parameters base
d on a retrospective case-control study,can be applied to other chronic diseas
es.