Fundamentals of epidemiology - an evolving text - Are you looking ...

More documents

Recommendations

Info

Dimensions in the training of an epidemiologist I. Epidemiologic perspective 1. Public health aspects: -- History of epidemiology, epidemiology as a public health science, clinical and public policy implications. 2. Scientific aspects: -- Problem conceptualization, philosophy of inference, study designs, interpretation of data, concepts of bias and multicausality. II. Measurement and analysis: Measures of disease frequency and extent, study designs and strategies, control of sources of error, statistical inference, data analysis and interpretation. III. Weighing epidemiologic evidence: Critical reading and synthesizing of information. IV. Proposal development: Specification of research hypotheses, study populations, measurement tools, analysis strategies; human subjects protection; “grantsmanship”. V. Study design and execution: Protocol development, subject recruitment, instrumentation, data collection, quality control, reporting and communications collaboration and working with oversight bodies, presentation of findings. VI. Data management: Manipulation and analysis of data using computers and statistical software packages. VII. Substantive knowledge: General background in health-related sciences and multidisciplinary understanding of specific areas of research. VIII. Epidemiologist roles: Development of skills for teaching, consultation, review of proposals and manuscripts, participation in professional meetings, leadership of multidisciplinary research teams, and continuing professional development. (Used for a number of years by the UNC Department of Epidemiology as an outline of areas of required competencies) _____________________________________________________________________________________________ www.epidemiolog.net, © Victor J. Schoenbach 1999, 2000 rev. 9/26/1999, 8/20/2000, 3/9/2001 1. Epidemiology Definition, functions, and characteristics - 16
2. An evolving historical perspective The evolution of epidemiology into a science of the distribution of disease in populations and evaluation of interventions for disease prevention and therapy. Propositions from studying history of epidemiology 1. Life has not always been the way it is in the developed countries today. 2. Scientific understanding of disease and the factors that affect it is largely a product of the last 150 years, with very rapid advances in the last half-century.. 3. Epidemiologic studies have not always been like ______ (insert the name of your favorite epidemiologic study). 4. There are many histories of epidemiology − History of health and disease − History of ideas and concepts − History of methods − History of knowledge gained through these concepts and methods − History of teachers and students − History of organizations and actions A brief history of public health Community attempts to prevent and limit the spread of disease go back to antiquity. For example, religious traditions against eating pork and shellfish reflect the special hazards of eating those foods when inadequately preserved or prepared. As often happens in public health, even without an understanding of the underlying etiology, effective preventive measures can be taken. Successes in prevention reinforce the concept that disease can be prevented through human action other than prayers and sacrifices to the gods, which in turn encourages additional attempts at prevention. By the 1600's, the practices of isolation and quarantine had begun to be employed to prevent the spread of certain diseases; by the 1800's these practices had become common in the American colonies. Methods of smallpox inoculation also began to be used and apparently mitigated some epidemics, even before Edward Jenner's introduction of a safe vaccine based on cowpox virus. __________ Acknowledgement: The following material draws heavily on lectures at the UNC Department of Epidemiology by Drs. Abraham Lilienfeld (1984) and Joellen Schildkraut (1989, 1990, 1991). _____________________________________________________________________________________________ www.epidemiolog.net, © Victor J. Schoenbach 1999, 2000 2. Historical perspective - 17 rev. 8/17/1999, 8/21/2000, 3/9/2001
Page 1 and 2: Understanding the Fundamentals of E
Page 3 and 4: Preface Introductory epidemiology c
Page 5 and 6: Table of Contents Chapter (in Acrob
Page 7 and 8: 1. Introduction Definition, charact
Page 9 and 10: problems (Schwartz and Carpenter, 1
Page 11 and 12: HIV - a new or newly-recognized vir
Page 13 and 14: d = principal investigator's degree
Page 15 and 16: Analytic studies typically involve
Page 17 and 18: Observational sciences especially a
Page 19: McGavran EG. What is public health?
Page 23 and 24: Time line for the history of public
Page 25 and 26: as a one-room bacteriology laborato
Page 27 and 28: − Health surveys − Research fun
Page 29 and 30: his scientific peers of the correct
Page 31 and 32: Kuhn, Thomas S. The structure of sc
Page 33 and 34: 3. Studying populations - basic dem
Page 35 and 36: The demographic balancing equation
Page 37 and 38: ates. Similarly, bulges in the repr
Page 39 and 40: When the baby boom cohort retires 1
Page 41 and 42: The pervasiveness, strength, viciou
Page 43 and 44: Total fertility rate (TFR) Standard
Page 45 and 46: Life expectancy The technique, of u
Page 47 and 48: Excerpt from the U.S. 1993 abridged
Page 49 and 50: At the other end of the life table,
Page 51 and 52: to change, hopefully to decline. If
Page 53 and 54: Thought question: Professors typica
Page 55 and 56: Descriptive studies and surveillanc
Page 57 and 58: 4. The Phenomenon of Disease Concep
Page 59 and 60: Classification is the foundation As
Page 61 and 62: As we gain more sophisticated under
Page 63 and 64: To recapitulate the above discussio
Page 65 and 66: Classifying cause of death Since mo
Page 67 and 68: Laboratory and other objectively me
Page 69 and 70: distribution but are not diabetic (
Page 71 and 72:
program in the general population,
Page 73 and 74:
Infectious disease Incubation "time
Page 75 and 76:
the direct effects of carcinogenic
Page 77 and 78:
patient outcomes. The design and ev
Page 79 and 80:
detected by screening will appear b
Page 81 and 82:
Morrison, Alan S. Screening in chro
Page 83 and 84:
e relevant. Compromises might be fo
Page 85 and 86:
educational attainment by the numbe
Page 87 and 88:
Distributions - the fuller picture
Page 89 and 90:
Per capita income: Should health ca
Page 91 and 92:
Deaths of children < 1 year of age
Page 93 and 94:
No. of persons with senile dementia
Page 95 and 96:
Incidence Immigration Relationship
Page 97 and 98:
Examples: arthritis, cholelithiasis
Page 99 and 100:
Cohort - entrance into the populati
Page 101 and 102:
Sample calculation: 200 people free
Page 103 and 104:
• The units of time must be state
Page 105 and 106:
Calculation of person-time in a coh
Page 107 and 108:
Original Calculation of person-time
Page 109 and 110:
average risk for a member of the co
Page 111 and 112:
However: Nurses: 601 cases/89,538 w
Page 113 and 114:
treated. If the rate at which patie
Page 115 and 116:
Incidence and prevalence in a popul
Page 117 and 118:
small prevalence. So if the seropre
Page 119 and 120:
Appendix on weighted averages Becau
Page 121 and 122:
Appendix on exponents and logarithm
Page 123 and 124:
and epidemiology. The notation ln(x
Page 125 and 126:
3. For the following hypothetical d
Page 127 and 128:
c. ________________ 6 fatalities /
Page 129 and 130:
Measuring disease and exposure - As
Page 131 and 132:
5. a. Flow Diagram Population of si
Page 133 and 134:
New cases Q ID = ------------------
Page 135 and 136:
Overview 6. Standardization of rate
Page 137 and 138:
differences between the groups in f
Page 139 and 140:
∑ (stratum-specific rates × stan
Page 141 and 142:
especially important when comparing
Page 143 and 144:
copy this cell to the other columns
Page 145 and 146:
where the RR k are the stratum-spec
Page 147 and 148:
and the observed number of deaths i
Page 149 and 150:
Directly standardized rate for A =
Page 151 and 152:
5,022 Indirectly standardized = —
Page 153 and 154:
the situation among older age group
Page 155 and 156:
Appendix on Standardized Mortality
Page 157 and 158:
across age strata, then the damage
Page 159 and 160:
d. How would you feel about the con
Page 161 and 162:
Cases per 100,000 Mean annual incid
Page 163 and 164:
d. dt Indirectly standardized rates
Page 165 and 166:
The second ingredient for an standa
Page 167 and 168:
Group US rate /100,000 County pop.
Page 169 and 170:
Crude rates are comparable because
Page 171 and 172:
The “Big Picture” 7. Relating r
Page 173 and 174:
CHD and oral contraceptives in Brea
Page 175 and 176:
Some basic measures Before diving i
Page 177 and 178:
esophageal cancer? First, if we did
Page 179 and 180:
ought in here as well. When we cont
Page 181 and 182:
Exposure Disease Yes No Total Yes a
Page 183 and 184:
Relation of the odds ratio to the r
Page 185 and 186:
the case-control study estimates th
Page 187 and 188:
If the data had come from a cross-s
Page 189 and 190:
Excess risk = CIR - 1 = 1.0 (i.e.,
Page 191 and 192:
strength of the relationship (e.g.,
Page 193 and 194:
drawn on ordinary graph paper. Howe
Page 195 and 196:
Measures of Impact Concept Relative
Page 197 and 198:
Attributable risk proportion The AR
Page 199 and 200:
show that a large proportion (i.e.,
Page 201 and 202:
Incidence Diagrammatic representati
Page 203 and 204:
1. high frequency of disease 2. pow
Page 205 and 206:
With this diagram and notation we c
Page 207 and 208:
vaguer) “attributed to”. Incide
Page 209 and 210:
Summary There are three categories
Page 211 and 212:
CIR can be directly estimated if th
Page 213 and 214:
I0 I = ——————— 1 + P1
Page 215 and 216:
PD|e PE (RR - 1) PD|e PE (RR - 1) P
Page 217 and 218:
Walter, Stephen D. Choice of effect
Page 219 and 220:
a. Quit rates were measured as the
Page 221 and 222:
1. Definitions: Relating risk facto
Page 223 and 224:
There is an extremely strong associ
Page 225 and 226:
8. Analytic study designs The archi
Page 227 and 228:
Progression of types of studies In
Page 229 and 230:
Cross-sectional A cross-sectional s
Page 231 and 232:
Follow-up studies Along with case-c
Page 233 and 234:
of interest community-level influen
Page 235 and 236:
ecological approach may be the appr
Page 237 and 238:
What measures can be estimated from
Page 239 and 240:
Ability to estimate risk Ascertainm
Page 241 and 242:
individuals, this relationship does
Page 243 and 244:
Odds Ratio (OR) Exposure odds in ca
Page 245 and 246:
Simple randomization - assignment o
Page 247 and 248:
The relationship of an attribute to
Page 249 and 250:
(a / m1) / (b / m1) ad ORe (Exposur
Page 251 and 252:
Validity The validity of a case-con
Page 253 and 254:
Since the strategic advantage of th
Page 255 and 256:
c = f0N1 = number of exposed noncas
Page 257 and 258:
experienced the event under study.
Page 259 and 260:
approach, the odds ratio computatio
Page 261 and 262:
Exposure Yes No Total Cases A B M1
Page 263 and 264:
Miettinen, Olli S. The clinical tri
Page 265 and 266:
Mantel, N. and Haenszel, W.: Statis
Page 267 and 268:
Peto, R., Pike, M.C., Armitage, P.,
Page 269 and 270:
4. The authors state that "There we
Page 271 and 272:
5. In contrast to many earlier inve
Page 273 and 274:
In this study the use of hospitaliz
Page 275 and 276:
Follow-up must be longer for any gi
Page 277 and 278:
While acknowledging the very import
Page 279 and 280:
world. Even as late as the 1950's,
Page 281 and 282:
Causal Inference Direct observation
Page 283 and 284:
data on the effects of radiation on
Page 285 and 286:
But excluding an explanation based
Page 287 and 288:
Strength of the association Pronou
Page 289 and 290:
Experimental evidence Certain types
Page 291 and 292:
To issue an indictment, the grand j
Page 293 and 294:
10. Sources of error A systematic f
Page 295 and 296:
the data in the table. (Note that t
Page 297 and 298:
associations. So the same terms may
Page 299 and 300:
"Negative bias" - The observed meas
Page 301 and 302:
Classifying sources of bias Inspite
Page 303 and 304:
The study population is a subset of
Page 305 and 306:
This estimate of the prevalence of
Page 307 and 308:
unmatched control groups will presu
Page 309 and 310:
The odds ratio for this table is 3.
Page 311 and 312:
Selection bias in cohort studies Se
Page 313 and 314:
Information bias Information bias r
Page 315 and 316:
following chapter). The concept of
Page 317 and 318:
P h y s i c i a n A Comparison of d
Page 319 and 320:
item). Alpha values of 0.80 are are
Page 321 and 322:
Hypothetical scenario showing effec
Page 323 and 324:
presumably classified as informatio
Page 325 and 326:
Appendix 1 Formulas to see the effe
Page 327 and 328:
Algebraically, _ Pr(E|D') = — the
Page 329 and 330:
Gregorio, David L.; James R. Marsha
Page 331 and 332:
Sources of error - Assignment 1. Th
Page 333 and 334:
Alpha (α) is the probability by wh
Page 335 and 336:
Sources of error - Assignment solut
Page 337 and 338:
3. ii) Selective survival--if OC us
Page 339 and 340:
a. Flow diagram: 1960-1962 1967-196
Page 341 and 342:
Introduction 11. Multicausality: Co
Page 343 and 344:
estimate of what the outcomes would
Page 345 and 346:
A hypothetical example (with apolog
Page 347 and 348:
Questions to ask: There are many as
Page 349 and 350:
Confounding arises from unequal dis
Page 351 and 352:
CHD Incidence by Behavior Pattern a
Page 353 and 354:
In this way, the 1,129 Type A's wit
Page 355 and 356:
exposure. If we assume a baseline r
Page 357 and 358:
Further insight can be gained by co
Page 359 and 360:
Table 7a Colon cancer and drinking
Page 361 and 362:
atio of the observed number of expo
Page 363 and 364:
ural-specific exposure prevalences,
Page 365 and 366:
eing infected, and the more of them
Page 367 and 368:
studies deal with a delimited geogr
Page 369 and 370:
University of Massachusetts at Amhe
Page 371 and 372:
similarly distributed in cases and
Page 373 and 374:
"There have been many important ste
Page 375 and 376:
MAIN POINTS Confounding is a disto
Page 377 and 378:
Greenland, S. and Neutra R.: Contro
Page 379 and 380:
Appendix The following discussion i
Page 381 and 382:
i n1i wi = ----- If exposure is unr
Page 383 and 384:
Multicausality: Confounding - Assig
Page 385 and 386:
4. The following table, published i
Page 387 and 388:
1. Multicausality: Confounding - As
Page 389 and 390:
link between HCS and MI to ensure t
Page 391 and 392:
Multicausality 12. Multicausality:
Page 393 and 394:
Interdependent effects The precedin
Page 395 and 396:
epidemiologic data. Put simply, the
Page 397 and 398:
BP 0 These two lines are parallel;
Page 399 and 400:
ean intake and B representing genet
Page 401 and 402:
Fourth, proportion of disease attri
Page 403 and 404:
Incidence and incidence ratios of f
Page 405 and 406:
% with evidence of impaired red blo
Page 407 and 408:
cancer rate among the light smokers
Page 409 and 410:
Age standardized lung cancer rates
Page 411 and 412:
R11 - R00 = R10 - R00 + R01 - R00 (
Page 413 and 414:
and R111 = R100 × R010 × R001 / (
Page 415 and 416:
R110 = R100 × R010 / R000 (multipl
Page 417 and 418:
Risk ln(risk) 0.19 -1.67 0.13 -2.01
Page 419 and 420:
have clinical or public health sign
Page 421 and 422:
Baseline Characteristics Associated
Page 423 and 424:
Weiss, Noel S. Accounting for the m
Page 425 and 426:
A. Synergism apparently exists in t
Page 427 and 428:
5. Walker (International Journal of
Page 429 and 430:
(or equivalently, the rate for pers
Page 431 and 432:
impact. Synergism in this sense imp
Page 433 and 434:
13. Multicausality - analysis appro
Page 435 and 436:
very far on instruments but needs t
Page 437 and 438:
Summarizing the relationships Often
Page 439 and 440:
We saw in the chapter on effect mod
Page 441 and 442:
So the multiplicative model for joi
Page 443 and 444:
misleading summary. These circumsta
Page 445 and 446:
w1 CID1 + w2 CID2 Summary CID = —
Page 447 and 448:
c d bcn0 + adn1 Var(ln(CIR)) ≈
Page 449 and 450:
Kleinbaum, Kupper, and Morgenstern;
Page 451 and 452:
1 + 1 + 1 3 ORMH = —————
Page 453 and 454:
Risk = R00 + Obesity effect Obesity
Page 455 and 456:
two (or more) factors that "interac
Page 457 and 458:
OR = exp(β3) β3 is the difference
Page 459 and 460:
Other regression models [Optional f
Page 461 and 462:
Know the epidemiologic meaning of t
Page 463 and 464:
14. Data analysis and interpretatio
Page 465 and 466:
and questionable data. Monitoring p
Page 467 and 468:
presence of particular proteins (e.
Page 469 and 470:
4. Count - the number of entities,
Page 471 and 472:
Algorithms - A procedure that uses
Page 473 and 474:
Imputation causes the least distort
Page 475 and 476:
Prenatal exposure to diethylstilbes
Page 477 and 478:
Possible outcomes (Colors of pairs
Page 479 and 480:
In a strict decision-making mode, t
Page 481 and 482:
C = area between -z and +z D = area
Page 483 and 484:
two-tailed 5% significance level).
Page 485 and 486:
If the p-value is not small (i.e.,
Page 487 and 488:
OR we observed was greater than 1.0
Page 489 and 490:
Large trials (e.g., 250 deaths) Tru
Page 491 and 492:
The concept behind the confidence i
Page 493 and 494:
Formulas for confidence intervals f
Page 495 and 496:
Woolf SH, Battista RN, Anderson GM,
Page 497 and 498:
Stallones, Reuel A. The use and abu
Page 499 and 500:
C - Clustered observations, which d
Page 501 and 502:
For example, suppose that n, the sa
Page 503 and 504:
* See previous table Margin of erro
Page 505 and 506:
Even in randomized trials problems
Page 507 and 508:
Pr(H|T) is therefore a function of
Page 509 and 510:
That information gives you pause. C
Page 511 and 512:
Part II 1. In table 2 from the Rose
Page 513 and 514:
If the OR for Current OC use were t
Page 515 and 516:
6. There are no interaction (produc
Page 517 and 518:
24. Submit manuscript from previous
Page 519 and 520:
Cooperative agreement A cooperative
Page 521 and 522:
Assurances by the applicant and her
Page 523 and 524:
The new element is the grant award.
Page 525 and 526:
1. What needs to be done? 2. How lo
Page 527 and 528:
Postaward Final report What to do
Page 529 and 530:
2. maintain close supervision of th
Page 531 and 532:
Publication Epidemiology Working Gr
Page 533 and 534:
Weed, Douglas L. Between science an
Page 535 and 536:
1.3 Specific Objectives of Data Man
Page 537 and 538:
engender a basic conflict between t
Page 539 and 540:
1.4.2 Integration Integrate the dat
Page 541 and 542:
2. Consistency of the protocol's im
Page 543 and 544:
2. Coding of a sample of data forms
Page 545 and 546:
2.3 The steps in the Editing proces
Page 547 and 548:
Although it is rare to use more tha
Page 549 and 550:
The first stages of data analysis s
Page 551 and 552:
have less power to exclude the stat
Page 553 and 554:
greater nonresponse by heavy drinke
Page 555 and 556:
Imputation is then carried out as f
Page 557 and 558:
(normal) distribution in the popula
Page 559 and 560:
Appendix **************************
Page 561 and 562:
___________________________________
Page 563 and 564:
Public health approach The public h
Page 565 and 566:
health, particularly at the world l
Page 567 and 568:
Causative or preventive agents Scur
Page 569 and 570:
Human behavior is also biology The
Page 571 and 572:
crowded, under-resourced urbanized
Page 573 and 574:
Bibliography Annas, George J.; Leon
Page 575 and 576:
18. Overview and Conclusion A look
Page 577 and 578:
Epidemiology is expanding In recen
Page 579 and 580:
More reliance on "soft money" - res
Page 581 and 582:
__________ This is not a new challe
show all

Fundamentals of epidemiology - an evolving text - Are you looking ...

Create successful ePaper yourself

Delete template?

Save as template?