True association? Errors in epidemiological studies Tipe tipe dasar Error 1. Random (chance) Error 2. Systematic Error/Sistematik error 2.1. Bias 2.2. Confounding 1. Error yang menyebabkan nilai-nilai yang diamati berbeda dari nilai yang sebenarnya. 2. Terjadi akibat variabilitas sample atau fluktuasi random dari kejadian yang diharapkan. 3. Error jenis ini dapat muncul pada setiap saat kita melakukan random sample dan membuat kesimpulan dari masing-masing populasi. 4. Hal ini tidak dapat dihindari, tidak dapat dikontrol dan dapat terjadi pada masing-masing arah/direksi. 5. Besarnya error dapat dibatasi dengan 5.1. Jumlah sample yang adekuat 5.2. Pengambilan sample yang tepat(Probability Sampling)
Random Error Random ERROR 1. Dapat terjadi setiap saat dalam penelitian
2. Dapat diatasi dengan :
2.1.Penghitungangan sampel minimal yang tepat dengan menggunakan perhitungan Probability Sample Size
2.2.Probability Random Sampling yang tepat = Pengambilan Sampel dari populasi memakai metoda Probability Random Sampling
Definition of bias Any systematic error in the design or conduct of an epidemiological study resulting in a conclusion which is different from the truth
Bias Yaitu bergerak menjauhi kebenaran atau harapan Penting diperhatikan dalam penelitian karena potensi efek pada makna dari hasil penelitian Faktor yang berkaitan dengan bias: Peneliti Alat pengumpul data Subyek penelitian Sampel Data Uji statistik Main sources of bias 1. Selection bias 2. Information bias Sumber-sumber bias 1. Proses seleksi atau partisipasi subyek ( bias seleksi) 2. Proses pengumpulan data ( bias informasi) 3. Tercampurnya efek pajanan utama dengan efek faktor risiko eksternal lainnya ( kerancuan/ confounding) Selection bias Two main reasons: Selection of study subjects Factors affecting study participation
association between exposure and disease differs between those who participate and those who dont 3 karakteristik penting bias seleksi 1. Terjadi ketika menggunakan kriteria yang berbeda dalam prosedur seleksi subyek 2. Besar dan arahnya seringkali tidak dapat diperkirakan 3. Bias ini, sekali terjadi tidak dapat dikendalikan, melainkan hanya dapat dicegah. Types of selection bias Sampling bias Ascertainment bias referral, admission Diagnostic/surveillance Participation bias self-selection (volunteerism) non-response, refusal survival Selection bias in case-control studies Selection of controls How representative are hospitalised trauma patients of the population which gave rise to the cases?
Cases liver cirrhosis Controls A trauma ward Heavy alcohol use 80 40 Light/no alcohol use 20 60
OR = 6 Estimate association of alcohol intake and cirrhosis Selection of controls
Cases liver cirrhosis Controls A trauma ward Controls B non-trauma Heavy alcohol use 80 40 10 Light/no alcohol use 20 60 90
OR = 6 OR = 36 Higher proportion of controls drinking alcohol in trauma ward than non-trauma ward a b c d Some worked examples Work in pairs In 2 minutes: Identify the reason for bias How will it effect your study estimate? Discuss strategies to minimise the bias Oral contraceptive and uterine cancer OC use breakthrough bleeding increased chance of testing & detecting uterine cancer
Cases uterine cancer Controls Takes oral contraceptives a b Does not take oral contraceptives c d
You are aware OC use can cause breakthrough bleeding
Overestimation of a overestimation of OR Diagnostic bias a b c d Lung cancer cases exposed to asbestos not representative of lung cancer cases Asbestos and lung cancer
Cases admitted and diagnosed with lung cancer Controls from surgical wards Contact with asbestos a b No contact with asbestos c d
Overestimation of a overestimation of OR Admission bias a b c d Prof. Pulmo, head specialist respiratory referral unit, has 145 publications on asbestos/lung cancer Selection bias in cohort studies Prospective cohort study- Year 1 Smoker 90 910 1000
Non-smoker 10 990 1000 lung cancer yes no 9 1000 10
1000 90 RR Loss to follow up Year 2 Smoker 45 910 955
Non-smoker 10 990 1000 lung cancer yes no 4.7 1000 10
955 45 RR 50% of cases that smoked lost to follow up Minimising selection bias Clear definition of study population Explicit case, control and exposure definitions CC: Cases and controls from same population Same possibility of exposure Cohort: selection of exposed and non-exposed without knowing disease status Sources of bias 1. Selection bias 2. Information bias
Information bias During data collection Differences in measurement of exposure data between cases and controls of outcome data between exposed and unexposed Information bias 3 main types: Recall bias Interviewer bias Misclassification
Bias informasi merupakan error/kesalahan dalam mendapatkan data dari masing- masing subjek Kesalahan pengukuran, Recall bias, Interviewer bias dll.
Bias Informasi 1. Subjek dengan penyakit yang dibuat-buat mungkin lebih intensif untuk mengingat paparan yang diduga sebagai faktor dibandingkan orang yang sehat. 2. Hal ini sangat penting pada kasus-kasus malformasi kongenital. Ibu dari bayi yang malformasi, ternyata lebih keras untuk mengingat kembali abnormalitas atau beberapa macam obat yang ia gunakan selama hamil dibanding ibu-ibu dengan bayi yang normal. 3. Recall bias timbul dari fakta bahwa, meskipun angka paparan pada 2 kelompok secara aktual sama, penyelidikan yang intensif diantara kasus-kasus akan menghasilkan proporsi paparan yang tinggi, yang menghasilkan odd ratio yang tinggi
Recall Bias Mothers of children with malformations remember past exposures better than mothers with healthy children Recall bias Mothers of Children with malformation Controls Took tobacco, alcohol, drugs a b Did not take c d Cases remember exposure differently than controls e.g. risk of malformation Overestimation of a overestimation of OR Investigator may probe listeriosis cases about consumption of soft cheese (knows hypothesis) Interviewer bias Investigator asks cases and controls differently about exposure e.g: soft cheese and listeriosis Cases of listeriosis Controls Eats soft cheese a b Does not eat soft cheese c d Overestimation of a overestimation of OR Misclassification Measurement error leads to assigning wrong exposure or outcome category Exposure Outcome Misclassification Systematic error Missclassification of exposure DIFFERS between cases and controls Missclassification of outcome DIFFERS between exposed & nonexposed => Measure of association distorted in any direction
Misclassification 250 100 150 100 50 50 Nonexposed 150 50 100 Exposed Total Controls Cases OR = ad/bc = 2.0; RR = a/(a+b)/c/(c+d) = 1.3 True Classification 250 100 150 110 60 50 Nonexposed 140 40 100 Exposed Total Controls Cases Differential misclassification OR = ad/bc = 3.0; RR = a/(a+b)/c/(c+d) = 1.6 Misclassification 250 100 150 100 50 50 Nonexposed 150 50 100 Exposed Total Controls Cases OR = ad/bc = 2.0; RR = a/(a+b)/c/(c+d) = 1.3 True Classification 250 100 150 110 50 60 Nonexposed 140 50 90 Exposed Total Controls Cases Differential misclassification OR = ad/bc = 1.5; RR = a/(a+b)/c/(c+d) = 1.2 Minimising information bias Standardise measurement instruments questionnaires + train staff Administer instruments equally to cases and controls exposed / unexposed Use multiple sources of information Summary: Controls for Bias Choose study design to minimize the chance for bias Clear case and exposure definitions Define clear categories within groups (eg age groups) Set up strict guidelines for data collection Train interviewers Summary: Controls for Bias Direct measurement registries case records Optimise questionnaire Minimize loss to follow-up The epidemiologists role 1. Reduce error in your study design 2. Interpret studies with open eyes: Be aware of sources of study error Question whether they have been addressed Confounding merupakan penyimpangan pengaruh faktor resiko dimana terjadi tumpang tindih pengaruh dari faktor resiko yang diteliti dengan pengaruh dari faktor-faktor resiko yang lain. Rothman menyebutnya: confusion of effects
Confounding
(E) Pajanan/ faktor risiko utama
(D) Penyakit yang ingin diteliti (C) Faktor eksternal/ confounder potensial
Confounding
Faktor-faktor luar yang bertanggung jawab dengan terjadinya confounding kemudian disebut confounder atau confounding variable. Bias confounding ini sendiri juga dikenal mempunyai banyak nama lain, seperti susceptibility bias, specification bias, Simpsons paradox, spurious association, secondary association.
Confounding
Perbedaan lain dengan bias seleksi dan bias informasi adalah bahwa jika confounders dapat diidentifikasi dan diukur secara adekuat pada seluruh subyek penelitian, kita dapat mengendalikan- (to control) atau menyesuaikan (to adjust) efek yang terdistorsi pada tahap analisis data, sementara pada golongan bias seleksi dan informasi pengendalian semacam ini pada tahap analisis tidak dimungkinkan Warning! Chance and confounding can be evaluated quantitatively Bias is much more difficult to evaluate Minimise by design and conduct of study Increased sample size will not eliminate bias