-
Notifications
You must be signed in to change notification settings - Fork 5
/
lec02.tex
1280 lines (1164 loc) · 41.1 KB
/
lec02.tex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
\documentclass[table,10pt]{beamer}
\mode<presentation>{
%\usetheme{Goettingen}
\usetheme{Boadilla}
\usecolortheme{default}
}
\usepackage{CJK}
\usepackage{graphicx}
\usepackage{amsmath, amsopn}
\usepackage{xcolor}
\usepackage[english]{babel}
\usepackage[T1]{fontenc}
\usepackage[latin1]{inputenc}
\usepackage{enumerate}
\usepackage{multirow}
\usepackage{url}
\ifx\hypersetup\undefined
\AtbBeginDocument{%
\hypersetup{unicode=true,pdfusetitle,
bookmarks=true,bookmarksnumbered=false,bookmarksopen=false,
breaklinks=false,pdfborder={0 0 0},pdfborderstyle={},backref=false,colorlinks=false}
}
\else
\hypersetup{unicode=true,pdfusetitle,
bookmarks=true,bookmarksnumbered=false,bookmarksopen=false,
breaklinks=false,pdfborder={0 0 0},pdfborderstyle={},backref=false,colorlinks=false}
\fi
\usepackage{breakurl}
\usepackage{color}
\usepackage{times}
\usepackage{xcolor}
\usepackage{listings}
\lstset{
language=R,
keywordstyle=\color{blue!70}\bfseries,
basicstyle=\ttfamily,
commentstyle=\ttfamily,
showspaces=false,
showtabs=false,
frame=shadowbox,
rulesepcolor=\color{red!20!green!20!blue!20},
breaklines=true}
\setlength{\parskip}{.5em}
\title[BI476]{BI476: Biostatistics - Case Studies}
\subtitle[observation]{Lec02: Observational Studies}
\author[Maoying Wu]{Maoying,Wu ({\small ricket.woo@gmail.com})}
\institute[CBB] % (optional, but mostly needed)
{
\inst{}
Dept. of Bioinformatics \& Biostatistics\\
Shanghai Jiao Tong University
}
\date{Spring, 2018}
\AtBeginSection[]
{
\begin{frame}<beamer>{Next Section ...}
\tableofcontents[currentsection]
\end{frame}
}
\begin{document}
\begin{CJK*}{UTF8}{gbsn}
\frame{\titlepage}
\begin{frame}
\frametitle{Goals of the lecture}
\begin{itemize}[<+->]
\item To understand the cross-sectional study along its strength and weakness.
\item To distinguish from the concepts of prevalence and incidence.
\item To define the organization of a case-control study.
\item To calculate and interpret the odds ratio (OR) as well as the confidence interval.
\item To understand the strength and weakness of CCS.
\item To understand the concept of bias and confounding.
\item To define and organize a cohort study.
\item To calculate and interpret the relative risk or risk ratio (RR).
\item To understand the different biases, and know how to detect them.
\item To understand the difference between association and causation.
\end{itemize}
\end{frame}
\begin{frame}
\frametitle{Outline}
\tableofcontents
\end{frame}
\section{Some important concepts and notions}
\begin{frame}[t]
\frametitle{Measures of Disease Frequency}
A fundamental task in medical studies is to quantify or measure the \alert{occurrence of
a disease in a population}. Obtaining a measure of the disease occurrence is one of the
first step in understand the disease of interest.
\begin{description}[<+->]
\item[\alert{Ratios}]{$\frac{a}{b}$, where $a$ is NOT part of $b$}
\item[\alert{Proportions}]{$\frac{a}{b}$, where $a$ is INCLUDED in $b$}
\item[\alert{Rates}]{$\frac{a}{b}$, where $a$ is the number of affected in a given time
interval, while $b$ is the population at risk over the same interval.}
\end{description}
\end{frame}
\begin{frame}[t]
\frametitle{Prevalence}
\textbf{Proportion of cases in the population at a given time}, indicating \alert{how widespread the disease is.}
$$
\textrm{Prevalence} = \frac{\mbox{\#cases observed at time } t}{\mbox{\#individuals at time} t}
$$
\begin{itemize}
\item An outbreak of diarrhea on a cruise ship in a day.
\item 8 persons out of 86 on the ship are exhibiting signs of diarrhea.
\item The prevalence of diarrhea at this particular time is $8/86=0.092$
\end{itemize}
\end{frame}
\begin{frame}[t]
\frametitle{Incidence}
\begin{itemize}
\item Cumulative incidence rate (CIR)
$$
\textrm{CIR} = \frac{\mbox{\#newly disease cases in a time interval}}{\mbox{\#individuals-at-risk}}
$$
\begin{itemize}
\item In the cruise ship example
\item 12 persons developed diarrhoeal disease after 5 days of the outbreak.
\item The indicidence rate was therefore $12/86=0.14$
\end{itemize}
\item Incidence density rate (IDR)
\begin{itemize}
\item \textbf{rate of occurrence of new cases}
\item conveys information about the \alert{risk of
contracting the disease}.
\end{itemize}
$$
\textrm{IDR} = \frac{\mbox{\#individuals of newly disease}}{\mbox{total time for all disease-free individuals-at-risk}}
$$
\end{itemize}
\end{frame}
\begin{frame}
\frametitle{Outcome and Exposure}
A good research statement of a medical problem is often simple, considering one
exposure and one outcome variable (e.g. organic solvents and brain tumor).
\begin{itemize}
\item Primary outcome
\item Secondary outcome
\end{itemize}
Research questions ask about the associations between exposures and outcomes.
\begin{quote}
Is proximity of the residents to the coke-works site associated with hospital
admission of children for repiratory problems?
\end{quote}
A hypothesis is then formulated to predict the result.
\begin{quote}
The risk of hospital admission for respiratory problems for children increases
with proximity to the coke-works site.
\end{quote}
\end{frame}
\begin{frame}[t]
\frametitle{Epidemiological Studies: Categories}
\begin{itemize}
\item Descriptive Studies
\begin{itemize}
\item Correlational (Ecological) studies
\item Case reports; Case series
\item \alert{Cross-sectional studies}
\end{itemize}
\item Analytic Studies
\begin{itemize}
\item Case-cohort study
\item Cohort study
\item Intervention study
\end{itemize}
\item Time-dependent
\begin{itemize}
\item Retrospective study
\item Prospective study
\end{itemize}
\end{itemize}
\end{frame}
\begin{frame}[t]
\frametitle{Ecological Fallacy}
\framesubtitle{Higher fat intake causes breast cancer?}
\begin{figure}
\includegraphics[width=0.50\textwidth]{../images/ecological_fallacies.png}
\end{figure}
\begin{itemize}
\item<1-> Ecological fallacy is an error in reasoning, usually based on
mistaken assumptions.
\item<2-> The ecological fallacy occurs when you \alert{make conclusions about
individuals based only on analyses of group data}.
\item<3-> In this fat intake example, we do not know individuals having
breast cancer were actually consumed high amount of fat.
\end{itemize}
\end{frame}
\section{Case Reports}
\begin{frame}[t]
\frametitle{Case Report: Example}
{\large \alert{Haemorrhagic-fever-like changes and normal chest radiograph in a doctor with SARS.}\\
\small{Wu E., et al. Lancet 2013; 361(9368): 1520-1521.}}
This case report is written by the patient, a physician who contracted SARS,
and his colleague who treated him, during the 2003 outbreak of SARS in Hong Kong.
They describe how the disease progressed in Dr. Wu and based on Dr. Wu’s case,
advised that a chest CT showed hidden pneumonic changes and facilitate a rapid
diagnosis.
{\large \alert{Kaposi's sarcoma in homosexual men-a report of eight cases.}}\\
\small{Hymes KB. et al. Lancet 1981; 2(8247): 598-600.}
This case report was published by eight physicians in New York city who had
unexpectedly seen eight male patients with Kaposi’s sarcoma (KS). Prior to this,
KS was very rare in the U.S. and occurred primarily in the lower extremities of
older patients. These cases were decades younger, had generalized KS, and a much
lower rate of survival. This was before the discovery of HIV or the use of the
term AIDS and this case report was one of the first published items about AIDS
patients.
\end{frame}
\begin{frame}[t]
\frametitle{Case Report: Definition}
Case report is an article that describes and interprets an individual case, often
written in the form of a detailed story.
Case reports often describe:
\begin{itemize}
\item Unique cases that cannot be explained by known diseases or syndromes
\item Cases that show an important variation of a disease or condition
\item Cases that show unexpected events that may yield new or useful information
\item Cases in which one patient has two or more unexpected diseases or disorders
\item the \alert{lowest-level} of evidence, but also the \alert{first-line} evidence.
\end{itemize}
A good case report will be clear about the importance of the observation being reported.
If multiple case reports show something similar, the next step might be a case-control
study to determine if there is a relationship between the relevant variables.
\end{frame}
\begin{frame}[t]
\frametitle{Case Report: Pros and Cons}
\uncover<1->{\begin{alertblock}{\center Advantages}
\begin{itemize}
\item Help the identification of new trends or diseases;
\item Help detect new drug side effects and potential uses (adverse or beneficial)
\item Educational - a way of sharing lessons learned
\item Identifies rare manifestations of a disease
\end{itemize}
\end{alertblock}}
\uncover<2->{\begin{alertblock}{\center Disadvantages}
\begin{itemize}
\item Cases may not be generalizable
\item Not based on systmatic studies
\item Causes or associations may have other explanations (confounding)
\item Can be seen as emphasizing the bizarre or focusing on misleading elements.
\end{itemize}
\end{alertblock}}
\end{frame}
\begin{frame}[t]
\frametitle{Case Series: Example}
\framesubtitle{TNF inhibitors treatment vs. chronic cutaneous sarcoidosis}
\begin{figure}
\includegraphics[width=0.85\textwidth]{../images/case_series_example.png}
\end{figure}
\end{frame}
\begin{frame}[t]
\frametitle{Case Series: Example}
\framesubtitle{Bullous pemphigoid}
{\large \alert{Clinical and Immunological Profiles of 14 Patients With Bullous
Pemphigoid Without IgG Autoantibodies to the BP180 NC16A Domain}}\\
\small{Kenta N. et al. JAMA Dermatol. 2018;154(3):347-350.}
This case series examines the association of nonreactivity of IgG to the
noncollagenous 16A domain using the enzyme-linked immunosorbent assay and
the chemiluminescent enzyme immunoassay and severity of disease course in
14 patients with bullous pemphigoid.
\begin{figure}
\includegraphics[width=.90\textwidth]{../images/bullous_pemphigoid.png}
\end{figure}
\end{frame}
\section{Cross-sectional study}
\begin{frame}[t]
\frametitle{Cross-sectional Studies: Definition}
\begin{itemize}
\item A cross-sectional study explores the disease and risk factor
patterns in a representive part of the population, in a
narrowly defined time period.
\item Primarily, this study provides information on \alert{prevalence}
of disease and risk factors.
\item It also can seek associations, generate and test hypotheses. And,
by repetition, cross-sectional study can be used to measure
changes.
\item Ideal cross-sectional study is of a geographically-defined,
representative sample of the population within a slice of
time and space.
\end{itemize}
\end{frame}
\begin{frame}[t]
\frametitle{Cross-Sectional Studies: Example}
\framesubtitle{Association between sleep and hypertension}
{\large \alert{Insomnia with objective short sleep duration is associated
with a high risk for hypertension.}}\\
\small{Vgontzas AN et al. Sleep 2009. 32:491-497.}
\begin{table}
\renewcommand\arraystretch{1.2}
\caption{Cross-sectional results of insomnia and hypertension}
\begin{tabular}{lccc}
\hline
& Hypertension(+) & Hypertension(-) & \\
\hline
Insomnia(+) & 121 & 78 & 199\\
Insomnia(-) & 837 & 705 & 1542\\
\hline
& 958 & 783 & 1741
\end{tabular}
\end{table}
Prevalence of hypertension: $958/1741 = 55\%$\\
Prevalence of insomnia: $199/1741 = 11\%$\\
Prevalence of hypertension among insomnia: $121/199 = 61\%$\\
Prevalence of hypertension among non-insomnia: $837/1542 = 54\%$\\
Prevalence ratio: $0.61/0.54 = 1.13$\\
Odds ratio: $\frac{121/78}{837/705} = $
\end{frame}
\begin{frame}[t]
\frametitle{Cross-Sectional Studies: Example}
\framesubtitle{The effect of sunless tanning products on tanning behaviors}
{\large \alert{A Cross-sectional Study Examining the Correlation Between
Sunless Tanning Product Use and Tanning Beliefs and Behaviors.}}\\
\small{Rachel E. et al. Arch Dermatol. 2012;148(4):448-454.}
\begin{figure}
\includegraphics[width=0.90\textwidth]{../images/sunless_tanning.jpg}
\end{figure}
\begin{itemize}
\item Time: May 30, 2007 - Dec 4, 2007
\item Location: The Emory University campus and surroundings in Atlanta,
Georgia
\item Participants: 415 community and university-affiliated women.
\end{itemize}
\end{frame}
\section{Case-control study}
\begin{frame}[t]
\frametitle{Case-Control Studies: Definition}
An epidemiological study in which a group of persons with the disease
of interest (\alert{case}) and a group of persons with similar features
to the case group but without the disease (\alert{control}) are selected
to compare the proportion of persons exposed to a risk factor of interest
(\alert{exposure}) in order to elucidate the causal relationship of the
risk factor of interest and the disease.
From \textbf{case series}, \textbf{personal experience}, and \textbf{others stuffs},
\begin{itemize}
\item Almost 99\% of patients suffered from condition $Y$ had a
history/evidence of exposure to $X$.
\item What is the problem in this case study?
\begin{itemize}
\item It will never prove a causal relationship due to lack of
control group.
\end{itemize}
\item It will generate a hypothesis testable using epidemiological methods:
\begin{itemize}
\item Persons with disease $Y$ were more likely to have been exposed
to factor $X$ comparing to persons without the disease, but
similar in other aspects.
\end{itemize}
\end{itemize}
That's the case-control study, a retrospective, observational study.
\end{frame}
\begin{frame}[t]
\frametitle{Case-control studies}
\uncover<1->{\begin{table}
\footnotesize
\caption{$2\times 2$ contingency table}
\begin{tabular}{lcc}
\hline
& Cases & Controls\\
\hline
Exposed(+) & $a$ & $b$\\
Exposed(-) & $c$ & $d$\\
\hline
\end{tabular}
\end{table}}
\uncover<2->{\begin{alertblock}{\center \small What we can answer}
\begin{itemize}
\item What are the odds that a case was exposed? \alert{$a/c$}
\item What are the odds that a control was exposed? \alert{$b/d$}
\item What is the odds ratio? \alert{$\frac{a/c}{b/d} = \frac{ad}{bc}$}
\end{itemize}
\alert{Note}: OR is calculated in different fashion in pair-match case-control
study.
\end{alertblock}}
\uncover<3->{\begin{alertblock}{\center \small What we CAN'T answer}
\begin{itemize}
\item the prevalence of disease in exposed and not-exposed.
\item the incidences of disease in exposed and not-exposed.
\item the relative risk to determine if there is an association
between the exposure and the disease.
\end{itemize}
\end{alertblock}}
\end{frame}
\begin{frame}[t]
\frametitle{Case-Control Studies: Computing Odds Ratio (OR)}
\begin{table}
\footnotesize
\caption{$2\times 2$ contingency table}
\begin{tabular}{lcc}
\hline
& Cases & Controls\\
\hline
Exposed(+) & $a$ & $b$\\
Exposed(-) & $c$ & $d$\\
\hline
\end{tabular}
\end{table}
\uncover<2->{\begin{alertblock}{\center Computing OR}
\begin{itemize}
\item The OR: $\widehat{\textrm{OR}} = \frac{a/c}{b/d} = \frac{ad}{bc}$
\item The standard error: $\mathrm{SE}\log \textrm{OR} = \sqrt{\frac{1}{a} + \frac{1}{b} + \frac{1}{c} + \frac{1}{d}}$
\item The 95\% confidence interval of $\log \textrm{OR}$: $\widehat{\log \textrm{OR}} \pm Z_{1-\alpha/2}^{-1} \mathrm{SE}\log \textrm{OR}$
\item $p$-value can be computed using either \texttt{Chi-square test} or \texttt{Fisher's exact test}.
\item In matched case-control study, \text{McNemar's test} is used to compute the $p$-value.
\end{itemize}
\end{alertblock}}
\end{frame}
\begin{frame}[t]
\frametitle{Case-Control Study Issues}
\framesubtitle{Bias and confounding}
In case-control studies, the inferred association can be introduced by the
subject selection process, the problem of data collection, and/or the other
factor(s) that are related to both exposure and the outcome under study.
Only when the two concerns - \alert{bias and confounding} are addressed, the
causality relationship can be derived.
\end{frame}
\begin{frame}[t]
\frametitle{Bias}
\begin{description}
\item[Bias]{Any systematic error in the \underline{design/subject recruitment},
\underline{data collection}, and/or \underline{data analysis} that
results in a mistaken estimation of the true exposure-outcome association.}
\item[Selection Bias]{Error due to systematic differences in characteristics
between those selected/non-selected into a study; or systematic differences
in which cases and controls, exposed and non-exposed subjects are selected
so that distorted association is observed.}
\end{description}
\uncover<1->{\begin{exampleblock}{Smoking-AMI Association}
\begin{itemize}
\item If hospitalized cases of AMI are selected as cases, and heavy/long-term smokers are more
likely to die outside of hospital (massive MI), the smoking-MI association would be
underestimated.
\item If hospitalized AMI and hostpitalized non-MI patients are selected as cases and controls,
and smokers are more likely to be hospitalized for other condition (e.g. pulmonary), the
smoking-AMI association would be underestimated.
\end{itemize}
\end{exampleblock}}
\end{frame}
\begin{frame}[t]
\frametitle{Selection Bias}
\begin{itemize}
\item Exposure has influence on the process of case assessment: the
exposure prevalence in cases is biased.
\item Exposure has influence on the process of control selection: the
exposure prevalence in controls is biased (e.g. use chronic
bronchitis patients as controls for study of smoking and CHD,
smoking and LC.)
\item \alert{Selective survival} and \alert{selective migration}: the
exposure prevalence in prevalent cases may be biased compared
to incident cases.
\end{itemize}
\uncover<2->{\begin{alertblock}{Case-control association is valid only when}
\begin{itemize}
\item Cases are selected as \alert{representative sample} of all people
with such disease.
\item Controls are selected as a representive sample of all people
without the disease in the same population as cases.
\end{itemize}
\end{alertblock}}
\end{frame}
\begin{frame}[t]
\frametitle{Information Bias}
\framesubtitle{Misclassification bias}
\begin{description}
\item[Information bias]{Systematic error due to inaccurate measurement and/or
classification of study variables. It can occur in the classification
of outcome, exposures, and covariates.}
\end{description}
\uncover<2->{\begin{exampleblock}{Example}
\begin{description}
\item[recall bias]{Lung cancer cases are more likely to recall their smoking
history (quantitatively or qualitatively), which will lead to over-estimation
of smoking-cancer relationship.}
\item[missing bias]{Overweighed persons are more likely to have missing data in
untrasound examination of carotid arteries.}
\item[report bias]{Females are more likely to under-report of their weight, and males
over-report.}
\end{description}
\end{exampleblock}}
\end{frame}
\begin{frame}[t]
\frametitle{Confounders (Confounding)}
\alert{Confounding} is a situation in which a measure of the association between
exposure and outcome is distorted because of the relationship between the exposure
and the other factor (the confounder) that influences the outcome under study.
Confounding is the most important concept because it impacts the \textbf{validity}
of an observational study.
\uncover<2->{\begin{alertblock}{Factor $X$ is a confounder only if it meets the 3 criteria}
\begin{itemize}
\item It is an risk factor of the outcome under study, independent of the risk
factor under study.
\item It is associated with the exposure under study.
\item It is not in the causal pathway between the exposure and the outcome.
\end{itemize}
\end{alertblock}}
\end{frame}
\begin{frame}[t]
\frametitle{How to avoid confounding association?}
\uncover<1->{\begin{alertblock}{In the design and conduct of study}
\begin{itemize}
\item \alert{Matching} (individual or group) by potential confounders - in
case-control studies.
\item Collect \alert{information} on potential confounders.
\end{itemize}
\end{alertblock}}
\uncover<2->{\begin{alertblock}{In the analysis of study}
\begin{itemize}
\item \alert{Conceptualize} your potential confounders.
\item \alert{Stratify} by potential confounders to identify major confounders.
\item Stratify to derive stratum-specific exposure-outcome association.
\item Statistical \alert{adjustment} of confounders.
\end{itemize}
\end{alertblock}}
\end{frame}
\begin{frame}[t]
\frametitle{Matched Case-Control Studies}
\begin{itemize}
\item<1-> Select controls who are identical (similar) to cases on potential confounders.
\begin{itemize}
\item<2-> \alert{Pair/Individual} match or \alert{frequency} match
\item<3-> Better control for confounding, especially when the distribution of a
confounder do not have much overlap between the cases and source
population (e.g., if cases of myocardial infaction tends to be older)
\item<4-> Analytical methods: \alert{Matched paired analysis} or \alert{conditional
logistic regression}
\end{itemize}
\end{itemize}
\uncover<5->{
\begin{table}
\caption{Pair-matched data}
\begin{tabular}{lcc}
\hline
\multirow{2}{*}{Cases} & \multicolumn{2}{c}{Controls} \\
\cline{2-3}
& Exposed & Non-exposed\\
\hline
Exposed & $a$ & \alert{$b$}\\
Non-exposed & \alert{$c$} & $d$\\
\hline
\end{tabular}
\end{table}}
\uncover<5->{\alert{How to compute the odds ratio (OR)?}}
\uncover<6->{$$
\textrm{OR} = \frac{b}{c}
$$}
\end{frame}
\begin{frame}[t]
\frametitle{Matched Paired Data Analsysis}
\begin{table}
\scriptsize
\caption{Pair-matched data}
\begin{tabular}{lcc}
\hline
\multirow{2}{*}{Cases} & \multicolumn{2}{c}{Controls} \\
\cline{2-3}
& Exposed & Non-exposed\\
\hline
Exposed & $a$ & \alert{$b$}\\
Non-exposed & \alert{$c$} & $d$\\
\hline
\end{tabular}
\end{table}
\uncover<2->{
\begin{alertblock}{\center \small Computing OR}
\begin{itemize}
\scriptsize
\item $a$ and $d$ are \alert{concordant pairs}; $b$ and $c$ are \alert{discordant pairs}.
\item The odds ratio: $\widehat{\mathrm{OR}} = \frac{b}{c}$
\item The standard error: $\mathrm{SE}_{\log \hat{\mathrm{OR}}} = \sqrt{\frac{1}{b} + \frac{1}{c}}$
\item The confidence interval of $\log \mathrm{OR}$:
$$
\log \widehat{\mathrm{OR}} \pm Z_{1-\alpha/2}\mathrm{SE}_{\log \hat{\mathrm{OR}}}
$$
\item The \alert{regular McNemar's Chi-square test}:
$$
\chi_{\textrm{McN}}^2 = \frac{(b-c)^2}{b+c} \sim \chi_{df=1}^2
$$
\item The \alert{continuity-correct McNemar's Chi-square test}:
$$
\chi_{\textrm{McN}}^2 = \frac{(|b-c|-1)^2}{b+c} \sim \chi_{df=1}^2
$$
\end{itemize}
\end{alertblock}}
\end{frame}
\begin{frame}[t]
\frametitle{A Paired Match Analysis}
\framesubtitle{estrogen and endometrial carcinoma}
{\large \alert{Association Of Exogenous Estrogen And Endometrial Carcinoma}}\\
\small{Donald C. Smith et al. NEJM 1975 Dec 4;293(23):1164-7.}
\begin{itemize}
\item Retrospective case-control study;
\item 317 patients with endometrium carcinoma
\item 317 matched controls with other gynecologic neoplasms
\end{itemize}
\begin{table}
\renewcommand\arraystretch{1.2}
\small
\caption{Paired study of association between estrogen and endometrial carcinoma}
\begin{tabular}{lccc}
\hline
& \multicolumn{2}{c}{Controls} & \\
\cline{2-3}
Cases & Exposed & Non-exposed & Total\\
\hline
Exposed & 39 & \alert{113} & 152\\
Non-exposed & \alert{15} & 150 & 165\\
\hline
Total & 54 & 263 & 317
\end{tabular}
\end{table}
\uncover<2->{
$$
\textrm{OR} = 113/15 = 7.5
$$
}
\end{frame}
\begin{frame}[t]
\frametitle{Estrogen and Endometrial Carcinoma}
\framesubtitle{Misuse of unmatched analysis}
{\large \alert{Association Of Exogenous Estrogen And Endometrial Carcinoma}}\\
\small{Donald C. Smith et al. NEJM 1975 Dec 4;293(23):1164-7.}
\begin{table}
\renewcommand\arraystretch{1.2}
\footnotesize
\caption{Unmatched study of association between estrogen and endometrial carcinoma}
\begin{tabular}{lccc}
\hline
& Cases & Controls & Total\\
\hline
Exposed & 152 & 54 & 206\\
Non-exposed & 165 & 263 & 428\\
\hline
Total & 317 & 317 & 634
\end{tabular}
\end{table}
\uncover<2->{
$$
\textrm{OR} = \frac{152/54}{165/263} = 4.5
$$
}
\uncover<3->{\begin{alertblock}{\center \small Conclusion}
\begin{itemize}
\item The unmatched $\textrm{OR}$ (4.5) is biased towards null compared
to the matched one (7.5).
\item The stronger the confounders, the biased the association.
\end{itemize}
\end{alertblock}}
\end{frame}
\begin{frame}[t]
\frametitle{Case-control Studies: Strengths and Limitations}
\uncover<1->{\begin{alertblock}{\center Advantages}
\begin{itemize}
\item Small sample size (rare disease)
\item Less time (disease with long induction and latent period)
\item Less expensive (small N, efficient for exposure that is
expensive to measure)
\end{itemize}
\end{alertblock}}
\uncover<2->{\begin{alertblock}{\center Disadvantages}
\begin{itemize}
\item Inefficient for rare exposure
\item Selection bias (Are cases representative? Do controls
represent the source population?)
\item Challenge in measurement of exposure (measure exposure
after the occurrence of disease)
\item Difficulty in determine temporality
\item Only one outcome is studied
\item Incidence of disease cannot be studied, though OR is
intended to estimate incidence ratio
\end{itemize}
\end{alertblock}}
\end{frame}
\begin{frame}[t]
\frametitle{Exercise: Case-control study}
\framesubtitle{Tabacco use and acoustic neuroma}
{\large \alert{Role of tobacco use in the etiology of acoustic neuroma}}\\
\small{Palmisano S et al. Am J Epidemiol. 2012 Jun 15;175(12):1243-51.}
\begin{itemize}
\item What kind of study? \alert{case-control study}
\item How many cases and controls? \alert{451 cases} and \alert{710
population-based controls.}
\item How did the author match the cases and controls?
\item Can you write down the $2\times 2$ contingency tables here?
\item How did the author analyze the data?
\item What was the effect size used in this article? and the confidence
interval?
\item Why did the author analyze the data in males and females, respectively?
\end{itemize}
\end{frame}
\section{Cohort Study}
\begin{frame}[t]
\frametitle{Cohort study}
\begin{itemize}
\item Observational study, with selection into study on basis of exposure
status at the beginning of the study
\item Either prospective study or retrospective study
\end{itemize}
\end{frame}
\begin{frame}
\frametitle{Flu Vaccine Study}
\framesubtitle{An epidemiology case study}
{\large \alert{Influenza Vaccination and Reduction in Hospitalizations
for Cardiac Disease and Stroke among the Elderly.}}\\
\small{Kristin Nichol et al.: NEJM 2003;348:1322-32.}
These investigators used the administrative data bases
of three large managed care organizations to study the
impact of vaccination in the elderly on hospitalization
and death. Administrative records were used to whether
subjects had received influenza vaccine and whether
they were hospitalized or died during the year of study.
\end{frame}
\begin{frame}[t]
\frametitle{Flu Vaccine Study}
\framesubtitle{Summarization table}
The table below summarizes findings during the \underline{1998-1999} flu season.
\begin{table}
\renewcommand\arraystretch{1.2}
\footnotesize
\caption{flu vaccine study data in 1998-1999}
\begin{tabular}{p{3cm}cc}
\hline
& Vaccinated & Unvaccinated\\
& $N=77,738$ & $N=62,217$ \\
\hline
Hospitalized due to pneumonia or influenza & 495 & 581\\
\hline
Hospitalized due to cardiac disease & 888 & 1026\\
\hline
Deaths & 943 & 1361\\
\hline
\end{tabular}
\end{table}
\alert{If the exposure is vacination and outcome of interest is death, how to assess the
association?}
\end{frame}
\begin{frame}[t]
\frametitle{Flu Vaccine Study}
\framesubtitle{Contingency table}
\begin{table}
\renewcommand\arraystretch{1.2}
\rowcolors{2}{gray!25}{white}
\scriptsize
\caption{flu vaccine study data in 1998-1999}
\begin{tabular}{p{3cm}cc}
\hline
& Vaccinated & Unvaccinated\\
& $N=77,738$ & $N=62,217$ \\
\hline
Hospitalized due to pneumonia or influenza & 495 & 581\\
\hline
Hospitalized due to cardiac disease & 888 & 1026\\
\hline
Deaths & 943 & 1361\\
\hline
\end{tabular}
\end{table}
\alert{If the exposure is vacination and outcome of interest is death, how to assess the
association?}
\begin{table}
\rowcolors{2}{white}{gray!25}
\renewcommand\arraystretch{1.2}
\small
\caption{flu vaccine and deaths in 1998-1999}
\begin{tabular}{lccc}
\hline
& Dead & Alive & \\
\hline
Vaccinated & 943 & 76,795 & 77,738\\
Non-vaccinated & 1361 & 60,856 & 62,217\\
\hline
\end{tabular}
\end{table}
\end{frame}
\begin{frame}[t]
\frametitle{Solutions}
\framesubtitle{Relative Risk (RR) and Risk Difference (RD)}
\begin{table}
\rowcolors{2}{white}{gray!25}
\renewcommand\arraystretch{1.2}
\scriptsize
\caption{A general $2 \times 2$ contingency table}
\begin{tabular}{lccc}
\hline
& Disease & Non-Disease & \\
\hline
Exposed & $a$ & $b$ & $r_1$\\
Non-exposed & $c$ & $d$ & $r_2$\\
\hline
\end{tabular}
\end{table}
\begin{columns}
\footnotesize
\begin{column}{0.45\textwidth}
\begin{block}{\center RR approach}
\textcolor{red}{$$
\textrm{RR} = \frac{a/r_1}{c/r_2}
$$}
\begin{itemize}
\item measuring the strength of the association.
\begin{itemize}
\scriptsize
\item $\textrm{RR}=1$ suggests no association.
\item $\textrm{RR} \to 1$ suggests weak association.
\item $|\textrm{RR}|>>1$ suggests a strong association.
\end{itemize}
\end{itemize}
\end{block}
\end{column}
\begin{column}{0.45\textwidth}
\begin{block}{\center RD approach}
\textcolor{red}{$$
\textrm{RD} = \frac{a}{r_1} - \frac{c}{r_2}
$$}
\begin{itemize}
\item a better measure of \alert{public health impact}.
\item How much impact would a prevention have?
\item How many people would benefit?
\end{itemize}
\end{block}
\end{column}
\end{columns}
\end{frame}
\begin{frame}[t]
\frametitle{Risk Ratio (RR)}
\framesubtitle{measuring the association between vaccination and death}
\begin{table}
\rowcolors{2}{white}{gray!25}
\renewcommand\arraystretch{1.2}
\small
\caption{flu vaccine and deaths in 1998-1999}
\begin{tabular}{lccc}
\hline
& Dead & Alive & \\
\hline
Vaccinated & 943 & 76,795 & 77,738\\
Non-vaccinated & 1361 & 60,856 & 62,217\\
\hline
\end{tabular}
\end{table}
$$\textrm{RR} = \frac{\textrm{CI}_e}{\textrm{CI}_u} = \frac{943/77738}{1361/62217} = 0.554$$
\begin{alertblock}{\center Conclusion}
\begin{itemize}
\item There is a strong association between flu vaccination and death?
\item Vaccination can protect ...?
\item How about the hospitalization due to two diseases?
\end{itemize}
\end{alertblock}
\end{frame}
\begin{frame}[t]
\frametitle{Risk Ratio: Confidence Interval}
\begin{table}
\rowcolors{2}{white}{gray!25}
\renewcommand\arraystretch{1.2}
\scriptsize
\caption{A general $2 \times 2$ contingency table}
\begin{tabular}{lccc}
\hline
& Disease & Non-Disease & \\
\hline
Exposed & $a$ & $b$ & $r_1$\\
Non-exposed & $c$ & $d$ & $r_2$\\
\hline
\end{tabular}
\end{table}
\uncover<2->{\begin{alertblock}{\center Confidence interval}
\begin{itemize}
\item $\hat{p_1} = \frac{a}{r_1}$; $\hat{p_0} = \frac{c}{r_2}$
\item $\widehat{\mathrm{RR}} = \frac{p_1}{p_0}$
\item $\mathrm{SE}_{\log \mathrm{RR}} = \sqrt{\frac{1-\hat{p_1}}{r_1 \hat{p_1}} +
\frac{1-\hat{p_0}}{r_2 \hat{p_0}}}$
\item The confidence interval of $\log \mathrm{RR}$: $\log \widehat{\mathrm{RR}} \pm
z_{1-\alpha/2} \mathrm{SE}_{\log \mathrm{RR}}$
\item Hypothesis testing: \alert{Chi-square test} or \alert{Fisher's exact test}
\end{itemize}
\end{alertblock}}
\end{frame}
\begin{frame}[t]
\frametitle{Risk Difference (RD)}
\framesubtitle{Attributable risk}
\begin{table}
\rowcolors{2}{white}{gray!25}
\renewcommand\arraystretch{1.2}
\small
\caption{flu vaccine and deaths in 1998-1999}
\begin{tabular}{lccc}
\hline
& Dead & Alive & \\
\hline
Vaccinated & 943 & 76,795 & 77,738\\
Non-vaccinated & 1361 & 60,856 & 62,217\\
\hline
\end{tabular}
\end{table}
$$\textrm{RD} = \textrm{CI}_e - \textrm{CI}_u = \frac{943}{77738} - \frac{1361}{62217} = -0.0097 = -97/100000$$ per year.
\begin{alertblock}{\center Conclusion}
\begin{itemize}
\item Flu vacinnation can reduce the death rate by 97 per 100,000 population per year.
\item Flu vaccination has a strong protective effect ....?
\item How about the hospitalization due to two diseases?
\end{itemize}
\end{alertblock}
\end{frame}
\begin{frame}[t]
\frametitle{Different conclusions for RR and RD}
\framesubtitle{Sometimes ...}
\begin{table}
\caption{Annual Mortality Per 100,000 (CI)}
\begin{tabular}{p{3cm}cc}
\hline
& LC & CHD\\
\hline
Smokers & 140 & 669\\