Editing Cohort study (section)

==Variations==
[[Image:ExplainingCaseControlSJW-en.svg|360px|Comparison of timelines of case-control and cohort studies. "OR" stands for "odds ratio" and "RR" stands for "[[relative risk]]".|thumb]]
The diagram indicates the starting point and direction of cohort and case-control studies.
In case-control studies the analysis proceeds from documented disease and investigations are made to arrive at the possible causes of the disease. In cohort studies the assessment starts with the putative cause of disease, and observations are made of the occurrence of disease relative to the hypothesized causal agent'''.<ref name="gfmer.ch">{{cite book |editor-first=Aldo |editor-last=Campana |first=O. |last=Meirik |title=Reproductive Health |chapter-url=http://www.gfmer.ch/Books/Reproductive_health/Cohort_and_case_control_studies.html |chapter=Cohort and Case-Control Studies |publisher=Geneva Foundation for Medical Education and Research}}</ref>'''

===Current and historical cohorts===
{{main|Retrospective cohort study}}

A current cohort study represents a true prospective study where the data concerning exposure are assembled prior to the occurrence of the fact to be studied, for instance a disease. An example of a current cohort study is the Oxford Family Planning Association Study in the United Kingdom, which aimed to provide a balanced view of the beneficial and harmful effects of different methods of contraception. This study has provided a large amount of information on the efficacy and safety of contraceptive methods, and in particular oral contraceptives (OCs), diaphragms and intrauterine device (IUDs).<ref>{{Cite journal|last1=Vessey|first1=M. P.|last2=Lawless|first2=M.|date=1984|title=The Oxford-Family Planning Association contraceptive study|journal=Clinics in Obstetrics and Gynaecology|volume=11|issue=3|pages=743–757|doi=10.1016/S0306-3356(21)00625-7|issn=0306-3356|pmid=6509857}}</ref>

In a historical cohort study the data concerning exposure and occurrence of a disease, births, a political attitude or any other categorical variable are collected after the events have taken place, and the subjects (those exposed and unexposed to the agent under study) are assembled from existing records or health care registers.

A "[[prospective cohort]]" defines the groups before the study is done, while historical studies, which are sometimes referred to as "[[retrospective cohort]]", define the grouping after the data is collected. Examples of a [[retrospective cohort]] are ''Long-Term Mortality after Gastric Bypass Surgery''<ref name="pmid17715409">{{cite journal |author=Adams TD |title=Long-term mortality after gastric bypass surgery |journal=N. Engl. J. Med. |volume=357 |issue=8 |pages=753–61 |year=2007 |pmid=17715409 |doi=10.1056/NEJMoa066603 |name-list-style=vanc|author2=Gress RE |author3=Smith SC |display-authors=3 |last4=Halverson |first4=R. Chad |last5=Simper |first5=Steven C. |last6=Rosamond |first6=Wayne D. |last7=Lamonte |first7=Michael J. |last8=Stroup |first8=Antoinette M. |last9=Hunt |first9=Steven C.|s2cid=8710295 |url=http://pdfs.semanticscholar.org/2908/f0ef315574fb7fa5a7941c97091a7fddd012.pdf |archive-url=https://web.archive.org/web/20190220132224/http://pdfs.semanticscholar.org/2908/f0ef315574fb7fa5a7941c97091a7fddd012.pdf |url-status=dead |archive-date=2019-02-20 }}</ref> and ''[[The Lothian Birth Cohort Studies]]''.<ref>{{cite web|url=http://www.psy.ed.ac.uk/research/lbc/LBC.html|title=The Lothian Birth Cohort Studies|access-date=8 May 2011|publisher=University of Edinburgh |archive-url=https://web.archive.org/web/20091022052040/http://www.psy.ed.ac.uk/research/lbc/LBC.html |archive-date=2009-10-22}}</ref>

Although historical studies are sometimes referred to as retrospective study, it a misnomer as the methodological principles of historical cohort studies and prospective studies are the same.<ref name="gfmer.ch" />

===Nested case-control study===
{{main|Nested case-control study}}
A nested case-control study is a case control nested inside of a cohort study. The procedure begins like a normal cohort study, however, as participants develop the outcome of interest they are selected as cases. Once the cases are identified, controls are selected and matched to each case. The process for selecting and matching cases is identical to a normal case control study. An example of a [[nested case-control study]] is ''Inflammatory markers and the risk of coronary heart disease in men and women'', which was a case control analyses extracted from the [[Framingham Heart Study]] cohort.<ref name="pmid15602020">{{cite journal |last1=Pai |first1=JK |title=Inflammatory markers and the risk of coronary heart disease in men and women |journal=N. Engl. J. Med. |volume=351 |issue=25 |pages=2599–2610 |year=2004 |pmid=15602020 |doi=10.1056/NEJMoa040967 |last2=Pischon |first2=T |last3=Ma |first3=J |display-authors=3 |last4=Manson |first4=Joann E. |last5=Hankinson |first5=Susan E. |last6=Joshipura |first6=Kaumudi |last7=Curhan |first7=Gary C. |last8=Rifai |first8=Nader |last9=Cannuscio |first9=Carolyn C.|s2cid=16142059 |doi-access=free }}</ref>

Nested case-controls have the advantage of reducing the number of participants that require details follow up or diagnostic testing to assess outcome or exposure status. However, this will also reduce the power of the study, when compared to larger cohort the study population is drawn from.

===Household panel survey===
Panel surveys are another important sub-type of [[longitudinal study]]. They differ from cohort studies by starting with representative cross-sectional samples, rather than cohorts defined by an event. Household panels draw representative samples of households and survey them, following all individuals through time on a usual annual basis. Examples include the US [[Panel Study of Income Dynamics]] (since 1968), the ''German'' [[Socio-Economic Panel]] (since 1984), the [[British Household Panel Survey]] (since 1991) and (since 2009) its successor [[Understanding Society: the UK Household Longitudinal Study]], the [[Household, Income and Labour Dynamics in Australia Survey]] (since 2001) and the European Community Household Panel (1994–2001).

===Cohort analysis in business===

For an example in business analysis, see [[cohort analysis]].

===AI for cohort study===
Conventionally, cohort studies require manual definitions of the common characteristics, which are time-consuming and labor-intensive, demanding extensive domain expertise. To address these limitations, researchers<ref>{{Cite journal|year=2024|title=CohortNet: Empowering Cohort Discovery for Interpretable Healthcare Analytics.|journal=Proc. VLDB Endow.|volume=17|issue=10|pages=22487–2500 |last1=Cai |first1=Qingpeng |last2=Zheng |first2=Kaiping |last3=Jagadish |first3=H. V. |last4=Ooi |first4=Beng Chin |last5=Yip |first5=James |doi=10.14778/3675034.3675041 |arxiv=2406.14015 }}.</ref> have increasingly explored the integration of AI technologies (e.g., CohortNet.<ref>{{cite web|title=GitHub - CohortNet|website=[[GitHub]] |url=https://github.com/KimballCai/CohortNet |access-date=2024-01-30}}</ref> COOL<ref>{{cite web|title=Introduction to COOL |url=https://www.comp.nus.edu.sg/~dbsystem/cool/ |access-date=2023-06-21}}</ref>) to automate the identification of cohorts with their definitions. For example, in healthcare, we can identify patients with a certain combination of feature conditions as a specific cohort, typically resulting in a similar outcome or end-point. Once the cohort is identified, we further learn the commonalities among the associated patients and obtain meaningful cohort representations. These AI-derived cohorts not only enhance the ability to evaluate new patients but also hold significant potential to accelerate medical research and discovery.