cohort study

cohort study

1. COHORT STUDY By Dr. Pragyan Paramita Parija DEPT. OF COMMUNITY MEDICINE VMMC & SAFDARJUNG HOSPITAL, NEWDELHI

2. Outline of the presentation: Epidemiology Concept Definition Design Types of cohort Measures of frequency Advantages Disadvantages Biases, confounding factor & synergism Examples

3. EPIDEMIOLOGY: THE STUDY OF THE DISTRIBUTION AND DETERMINANTS OF HEALTH RELATED STATES OR EVENTS IN SPECIFIED POPULATION AND APPLICATION OF THIS STUDY TO CONTROL HEALTH PROBLEMS. (JOHN M.LAST) **Leon Gordis.5 th edition:p.2

4. Determinants of health:

5. Types of epidemiology:

6. Concept of cohort Cohort is defined as a group of people who share a common characteristic or experience within a defined time period. TYPES OF COHORT: CLOSED/FIXED : Fixed group of persons followed from a certain point of time to defined end population OPEN/DYNAMIC : Persons may leave/enter the study at any point without compromising study design. * MC.Mohan Brian,p:167

7. CHOICE OF STUDY COHORTS: 1. They have undergone some unusual exposure/experience of which the effects are to be evaluated 2. Often come across some special resource that may facilitate ascertainment of their exposure, follow up or disease experience. ** Mc.Mohan Brian.p:172

8. Distinguish features of cohort study: Cohorts identified prior to appearance of disease under investigation Study group defined observed over a period of time to determine frequency of disease among them Cause Effect Cohort study which is usually undertaken to obtain additional evidence to refute/support existence of an association between suspected cause and disease.

9. Definition: It is also called as prospective cohort study or concurrent cohort study or longitudinal study or forward looking study. It is concurrent because investigator identifies the original population at the beginning of study and in effect, accompanies the subjects concurrently through calendar time until at which disease develops or does not develop.

10. Prospective study Retrospective cohort study Ambi -directional cohort study Types of cohort study

11. Defined population Exposed Not exposed Non randomization Disease No disease Disease No disease Prospective 2000 2010 2020 Retrospective 1987 1997 2007 Ambi 1987 2007 2017

12. The common strategy of cohort studies is to start with a reference population, some of whom have certain characteristics or attributes relevant to the study(exposed) with others who don’t have those characteristics(unexposed). Both groups should be free from conditions under consideration. Investigator accompanies the subjects concurrently through calendar time until the disease develops or not. The common strategy of cohort studies is to start with a reference population, some of whom have certain characteristics or attributes relevant to the study(exposed) with others who don’t have those characteristics(unexposed). Both groups should be free from conditions under consideration. Investigator accompanies the subjects concurrently through calendar time until the disease develops or not. **Leon Gordis,5 th edition.p182 Prospective cohort study

13. A historical cohort study depends upon availability of data or records that allow reconstruction of the exposure of cohorts to a suspected risk factor and follow up of their mortality or morbidity over time. Although the investigator was not present when exposure was first identified, he reconstructs exposed and unexposed populations from records and then proceeds as though had been present throughout study. Outcome is ascertained at the time of study. Retrospective / Historical cohort study

14. With this approach, exposure is ascertained from objective records in the past( as in a historical cohort study), and follow up and measurement of outcome into the future. E.g. Court & Doll-Brown study Cohort assembled in 1955 consisting of 13352 pts who had received large dose of radiation for Ankylosing spondylitis between 1934 -1954. Outcome evaluated was death from leukemia between 1935 & 1954. and found that death rate was higher in their cohort than general population. A prospective element was added, as cohort was followed to identify deaths occurring in subsequent years. Ambi directional study

15. Indication for cohort studies: There is good evidence of an association between exposure and disease( derived from clinical observation and supported by descriptive and case-control study ) Exposure –rare, incidence –high Attrition of study can be minimized. (follow up easy, cohort stable cooperative accessible ) Ample fund

16. General consideration: Cohort must be free from disease Both groups should be equally susceptible to the disease Both groups should be comparable in all aspect Diagnostic and eligibility criteria must be defined before hand Groups followed under the same identical condition over a period of time to determine outcome

17. Design: Selection of cohort Collection of data on the exposure Selection of comparsion population Follow up Analysis

18. Selection of cohort: A community cohort of specific age and sex An exposure cohort e.g. smokers, users of OCP, radiologists A birth cohort , e.g. school entrants An occupational cohort, e.g. miners A marriage cohort A diagnosed cohort or treated cohort, e.g. cases treated with radiotherapy, surgery. The usual procedure is to locate or identify the cohort which may be a total population in an area or sample thereof . If study start with exposed and non exposed, it can explore only specific exposure. If study start with defined population it can explore multiple exposure. * WHO Research methodology,2 nd edition.p:28

19. Data collection In a conventional cohort study, an initial cross sectional study is often performed to exclude persons with the outcome of interest(disease) and to identify the cohort that is free from the disease. Data collected from cohort members: interviews/mailed questionnaires Medical records Medical examination Environmental survey **WHO Research methodology

20. Selection of comparsion group : Internal Control Group Exposed and non-exposed in the same Study population (Framingham study) Minimise the differences between exposed and non-exposed External Control Group When information on degree of exposure is not available chose another group, another cohort (radiologists cohort with a cohort of ophthalmologist) General Population : If none of the above comparison is available than the mortality experience of the exposed group is compared with the mortality experience of the general population in the same geographic area as the exposed people. E.g. comparison of frequency of cancer among uranium mine workers with the rate in general population in same geographic area. Limitation: non availability of outcome data, difficulty in selecting study and control

21. Ideal comparsion : A group of thymic enlargement at similar stage with similar clinical profile who were not irradiated to rule out thymic enlargement themselves were responsible for any usual outcome. Rarely available in non experimental study On the basis of indication and contra indication: introduce confounding of the effects of treatment by its own indication/contraindication( confounding by indication ) ** Mc.Mohan Brian.p:188

22. Follow up:

23. 1. Cumulative incidence : It is based on the total population at risk, which was, at entry to the study, free of the disease under investigation. It is calculated for each stratum of exposure to the risk factor and is the ratio of number of new cases or events in a specified period of observation to the total population at risk during that period. Cumulative incidence is a proportion, not a rate and can vary from 0 to 1. i.e. no less than 0% and not more than 100% E.g. if 100 patients observed for 1 year ,5 of them developed hepatitis A during 1 yr period who are initially free of disease, then cumulative incidence=5/95 Measures of frequency

24. 2. Incidence density(person-time approach) It is an improvement over the conventional measure of incidence, because it takes into consideration both the number observed and the duration of observation for each individual.

25. Example : if 30 individuals were observed; 10 for 2years, 5 for 3years, and 15 for 4years,they would contribute 10* 2+5*3+15*4=95 person years of observation( denominator ) numerator – no. of new cases observed in these groups over the specified period of time. Person years do not represent the no. of persons If 400 person years then it can either represent 400 persons each observed for 1year or 40 persons for 10years

26. Drawbacks of incidence density: Exact time when the disease occurs cannot be ascertained Rate of disease development over time is not necessarily constant

27. DISEASE(OUTCOME) EXPOSURE PRESENT ABSENT TOTAL PRESENT a b a+b ABSENT c d c+d TOTAL a+c b+d a+b+c+d =N

28. Other frequency measures: I(exposed)=a/ a+b I(non exposed)=c/ c+d Relative Risk=I exposed/I non exposed Attributable risk(AR)=I exposed - I non exposed Attributable risk percent= [( a+b /N)(RR-1)]/[1+( a+b /N)(RR-1)] vi. Population attributable risk= Incidence among population – incidence among non exposed vii. Etiologic/attributable fraction=I exposed – I non exposed/ I exposed AR measures public health problem caused by risk factor. RR measures the strength of the association

29. HYPERTENSION SMOKING PRESENT ABSENT TOTAL PRESENT 120(a) 280(b) 400( a+b ) ABSENT 30(c) 570(d) 600( c+d ) TOTAL 150( a+c ) 850( b+d ) 1000(N)

30. I(exposed)=a/ a+b = 120/400 =0.3 I(non exposed)=c/ c+d = 30/600 = 0.05 Relative Risk=I exposed/I non exposed = 0.3/0.05 =6 Attributable risk(AR)=I exposed - I non exposed= 0.30- 0.05 =0.25 Etiologic fraction= Ie – In / Ie =1-1/RR = 0.30 – 0.05 / 0.30 =83.3%

31. RR=1 = No association between exposure and disease incidence rates are identical between groups RR=> 1 = Positive association exposed group has higher incidence than non-exposed group RR=< 1 = Negative association or protective effect. non-exposed group has higher incidence than exposed or exposed group has lower incidence than non-exposed e.g. RR 10% / 20% = 0.5 it would indicate that if one smokes, the risk of getting IHD is 10%; on the other hand if one does not smokes, the risk is 20%. Smoking thus reduces the risk of getting IHD by half. Interpretation of relative risk:

32. Chi square test:

33. E1= ( a+b )( a+c )/N =400* 150/1000=60 E2= ( a+b )( b+d )/N=400*850/1000=340 E3= ( c+d )( a+c )/N=600*150/1000=90 E4= ( c+d )( b+d )/N=600*850/1000=510 O1=a=120, O2=b=280, O3=c=30, O4=d=570 £ (O-E)²/E ={(120-60)²/60+(280-340)²/340+(30-90)²/90+(570-510)²/510} =60+10.58+40+7.05=117.63 >3.84 for degree of freedom 1(i.e . significant )

34. advantages

35. Disadvantages:

36. Attrition: loss of f/u; > 10-15%:more serious potential bias

37. Disadvantages of retrospective cohort study:

38. Biases in cohort study:

39. Confounding factor: Independently associated with exposure and outcome and unequally distributed among study and control group An association of confounding factor is always less strong than the association of the confounding factor with the disease under study and less strong than the association of this factor with the exposure under investigation.

40. confounding is evaluated and controlled for analysis

41. Synergism: It implies a certain factor that has a different effect in the presence of another factor rather than its absence. Joint effect of two factors would have been different rather than the individual factor. Additive model : more appealing and relevant Multiplicative model : flexible and convenient ( Mantel- haenzel method, logistic regression, proportion hazard model) ** Mc.Mohan Brian.p:214-5

42. E.g. In a study of male asbestos workers, cumulative incidence of lung CA among exposed who were smokers was much greater than sum of the risks associated with asbestos and smoking alone Risk ratio of lung CA in smokers compared to non smokers was 10 . Risk ratio with 20yrs asbestos exposure was found to be of same order magnitude Asbestos workers who smoked estimated to have 92 times risk of dying of lung CA than who neither smoked nor worked with asbestos. Here multiplicative model: no interaction Additive model: substantial interaction

43. A – PRESENCE OF ONE FACTOR A‾ – ABSENCE OF FACTOR B – PRESENCE OF ANOTHER FACTOR B‾ – ABSENCE OF ANOTHER FACTOR SYNERGY INDEX={RR(AB)—1}/{RR(A B‾ )+RR( A‾ B) —2} Additive model, S=1, Absence of interaction Higher value: positive interaction Lower value: negative interaction [ Antagonism ] ** Mc.Mohan Brian.p:214

44. THE FRAMINGHAM STUDY Best-known cohort studies is the Framingham Study of cardiovascular disease. Started in 1948 Framingham is a town in Massachusetts, about 20 miles from Boston which was town meeting form of Govt . people were well accustomed to group approach. Because community study of TB was done here for 6 yrs since 1917. Residents between 30 and 59 years of age were considered eligible for study. 1971 enrolled a second generation of participants. In April 2002 , a third generation was enrolled in the core study.

45. US PUBLIC HEALTH planned for this study in 1947 in cooperation with state and local health agencies. Study is focused on arteriosclerotic coronary artery disease and hypertension HYPOTHESIS : THESE DISEASES DON’T EACH HAVE A SINGLE CAUSE (AS IN INFECTICIOUS DISEASE) BUT RESULT OF MULTIPLE CAUSES WHICH WORK SLOWLY WITHIN INDIVIDUAL RESEARCH FRAMEWORK : BASED ON CLINICAL EXAMINATION, free of definite signs of disease considered as normal, observed till acquire disease. Search is made for factors in one group and not in other BY PRODUCT OF THIS INVESTIGATION: 1. Study the efficiency of various diagnostic procedures finding heart disease 2. Prevalence and Incidence of CVD

46. Ideally, various race, ethnicity, SES, environment should considered. A single area only when large population size available Birth to death cohort is best But expense and F/U not practicable to carry. So concluded that; A single area , approx. 6000 population for 20yrs . A town of 25000-50000 population can supply these no. of adults. A town is more desirable than a large city because community approach require secure full cooperation and coverage. Belief that arteriosclerotic disease and hypertension in white race is more within community variants than between community variants ** Dawber , T.R. etal 1951 Am. J Pub Hlth 41:279

47. Study subjects selection: Age range: very young; <20 yrs , very small population develop disease Very old: too large population develop disease To balance this, 30-59 yrs,10000 population was there as per age and sex distribution. 5000 was normal at the initial of study 1 st stratified by family size ,then precinct of residence then by address >20 yrs …Framingham listed all ppl If one family member was taken then all other members of same household also taken into consideration( if within eligible age range) Sampling ratio was 2/3. so 6000 taken then extra 10% for migration or refusal.. so 6600.

48. Hypothesis : Incidence of CHD increases with age Hypertension develop CHD Elevated cholesterol is associated with CHD Tobacco smoking and habitual use of alcohol increased CHD Increased physical activity a/w decreased incidence of CHD Increased Body weight increases incidence of CHD Diabetes increases incidence of CHD New coronary events were identified by examining the study population every 2 years and by daily surveillance of hospitalizations at the only hospital in Framingham.

49. Results: 1960s: Cigarette smoking, Increased cholesterol, elevated blood pressure, obesity increases risk of heart disease. Exercise decreases risk of heart disease. 1970s: Elevated blood pressure increases risk of stroke. Postmenopausal women risk of heart disease is increased compared with who are premenopausal. 1980s High levels of HDL cholesterol reduce risk of heart disease. 1990s : Elevated blood pressure can progress to heart failure. At 40 years of age, the lifetime risk for CHD is 50% for men and 33% for women.

50. 2000s “High normal blood pressure" increases risk of cardiovascular disease (high normal blood pressure is called pre hypertension in medicine; it is defined as a systolic pressure of 120–139 mm Hg and/or a diastolic pressure of 80–89 mm Hg). Lifetime risk of developing elevated blood pressure is 90%. Serum aldosterone levels predict risk of elevated blood pressure. Lifetime risk for obesity is approximately 50%.

51. It is recognized that breast cancer is more common who are older at the time of their 1 st pregnancy. 2 questions arised : Early 1 st pregnancy protects against breast cancer Both a delayed 1 st pregnancy and increased risk of breast cancer due to some third factor(hormonal abnormality) 2. COWAN’S RETROSPECTIVE COHORT STUDY OF BREAST CANCER:

52. 1978: Cowan and coworkers identified a population at John Hopkins Infertility Clinic in Baltimore, Maryland from 1945 to 1965. By definition all the subjects had a late age at 1 st pregnancy. In the course, their hormonal profiles were developed for each woman. The researchers were able to separate the women with hormonal abnormality from those who had another cause of infertility (tubal patency or husband’s low sperm count )

53. Patients of JHH infertility clinic (Delayed 1 st Pregnancy) Hormonal abnormalities No hormonal abnormalities Develop cancer Do not develop cancer Develop cancer Do not develop cancer 1945-1965 1978

54. The study found that, when the development of breast cancer was considered for the entire group, the incidence was 1.8 times greater in women with hormonal abnormalities than women without hormonal abnormalities. But finding was not statistically significant. When the occurrence of breast cancer was divided into categories of premenopausal and post menopausal incidence, women with hormonal abnormalities had 5.4 times greater risk of premenopausal occurrence of breast cancer. Results:

55. No difference was seen for postmenopausal occurrence of breast cancer. But it is not clear whether this lack of difference in incidence of postmenopausal breast cancer is true absence of difference or a small no. of women in this population had reached menopause at the time study was conducted.

56. Prognostic cohort study: It might influence prognosis after diagnosis and treatment Cases diagnosed / treated at a fixed time[ by medical/surgical/rehabilitation/psychological/vocational Such cases are not free of specified disease as like conventional cohort study(but should be free of outcome of interest ) Outcome measured: survival/cure/improvement/disability **WHO Research methodology,2 nd edition. p: 34

57. Sample size

58. Health research methodology , 2 nd edition Mc.Mohan Brian,1 st edition Oxford text book of public health,6thedition.p:468-483. Text book of epidemiology, Leone Gordis , 5 th edition Text book of preventive and social medicine, K. Park, 23 rd edition Basic epidemiology, R Bonita, 2 nd edition Dawber , T. etal (1951) Am J Pub.Hlth.41:279 Charan JK, Biswas T. How to calculate sample size for different study design in Medical research. Indian J Psychological Med.2013 Apr;35(2):121-6 REFERENCES: