KowalskiD pg. 1 Scientific Study Designs for research. What Research is ... the systematic process of collecting and analyzing information (data) in order to increase our understanding if phenomenon about which we are concerned or interested. Apa pentingnya penelitian? Contohnya untuk mengetahui kontra indikasi kb? Tidak boleh diberikan pada pendarahan pervaginam yang gak diketahui sebabnya/DisfUterinBleeding misalnya. Mengetahui amannya bagaimana? Karena SUDAH terbukti di penelitian. Seperti merokok, hasil penelitian menunjukkan ada efek sampingnya. Tanpa penelitian kita tidak akan tahu bahwa penyebab ca serviks adalah HPV dan karena penelitian juga maka ada vaksinnya. Jadi peran penelitian, ilmu makin maju dan kesejahteraan manusia meningkat. Penelitian itu mencari KEBENARAN. Untuk mencari kebenaran harus menggunakan metode yang benar, tata cara yang benar, dll. Kalau tidak penting tidak perlu diteliti, begitu pun variabel yang di tanyakan nantinya, yang di anggap penting saja. Research Characteristics 1. Originates with a question or problem. 2. Requires clear articulation of a goal. 3. Follows a spesifik Plan or prosedur (Scientifiec method) 4. Often divides main problem into subproblems. 5. Guided by spesific problem, question, or hypothesis. 6. Accepts certain critical assumtions. 7. Requires collection and interpretation of data. 8. Cyclical (helical) in nature. Karakter/ ciri-ciri penelitian : (Contohnya akan dilakukan penelitian tentang medical check up) Berasal dari sebuah pertanyaan atau problem. Contoh, seberapa banyak orang yang melakukan medical checkup? Ada Goal tujuannya. Apakah tujuannya hanya mengetahui presentasi saja? Atau cari penyebab atau pencegahannya. Scientific method. Pendekatan atau prosedur ilmiahnya seperti apa? Contoh prosedur ilmiah : Riset problem > penelusuran literatur > kerangka teoritisnya > kerangka konsepnya > pengumpulan data dan analisis > kesimpulannya apa. Divides main problem kedalam subproblem. Misalnya masalah medical checkup, fokus pada checkup ca serviks (pap smear) saja misalnya. Jadi bikin subproblem yang lebih spesifik dari main problem yang umum. Guided by spesific problem. Bikin hipotesisnya. Hypotesis, misalnya biaya mungkin menyebabkan kebiasaan medical checkup. Nanti di buktikan hipotesisnya benar atau tidak. Question problem, misalnya apakah ada hubungan dengan tingkat pendidikan atau pengetahuan dengan kepatuhan pemeriksaan pap smear (check up ca serviks)? Critical assumtions,contohnya asumsi saya ibu tidak punya biaya, alat gak ada, dll sehingga tidak melakukan pap smear. Nanti asumsi ini kemudian akan dibuktikan. Collection dan interpretasi data. Harus punya data, misalnya usia ibunya, pendidikan, pengetahuan, usia menikah dsb. Cyclical (helical) in nature. Secara siklus dalam alamiah. Menggambarkan secara teoritis masalah tersebut dan manfaatnya apa? Kuliah Pertama Metodologi Penelitian. Dr. Djap. BOLD : SLIDE KULIAH Regular : Catatan Rekomendasi bacaan ada di lampiran.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Scientific study designs for research: an overview Anita S Acharya*, Anupam Prakash** Aruna Nigam#, Pikee Saxena#
Symposium
Departments of *Community Medicine, **Medicine, and #Obstetrics & Gynaecology. Lady Hardinge Medical College, New Delhi-110001. India.Corresponding Author: Dr. Anita S Acharya, Associate Professor, Dept. of Community Medicine, Lady Hardinge Medical College, New Delhi-110001. India. E-mail: [email protected]: 11-08-2012 | Accepted: 13-08-2012 | Published Online: 26-08-2012This is an Open Access article distributed under the terms of the Creative Commons Attribution License (creativecommons.org/licenses/by/3.0)Conflict of interest: None declared Source of funding: Nil
Abstract
Research is a scientific, well planned methodical attempt to answer questions with valid data. “The rules that govern the process of collecting and arranging the data for analysis are called research designs”. They are broadly classified into “Observational” and “Experimental” study designs. In observational study design, the researcher simply observes and does not intervene in any way whereas in the latter, some kind of intervention/manipulation is done. Descriptive, observational study designs are useful for only generating hypothesis whereas, analytical, observational study designs are helpful for both generating and testing hypothesis. Randomised controlled trials are the “Gold Standard” for determining the strongest evidence for concluding causation. However, no study design is perfect. Each has its own inherent advantages and disadvantages. Depending on the type of research question, practicability, and resources in terms of manpower, money and time, the investigator has to choose the appropriate study design which will answer the research question in the most scientific manner. This article gives a brief overview of the various study designs commonly used in research.
Key words: Research design; cross-sectional; case-control; cohort; randomised controlled trial.
Research is derived from “re-search”. Search refers to “seek”, “investigate” or “explore” and in typical Hindi parlance it means “khoj”. Research may be a new search or an old thing being again investigated (re-search).
Once a research question is in place, what kind of study should be planned is important and this is primarily the focus of this article.
A study design is a specific plan or protocol for conducting the study, which allows the investigator to translate the conceptual hypothesis into an operational one. In other words “The rules that govern the process of collecting and arranging the data for analysis are called research designs” [1] which we shall be discussing in brief along with their advantages and dis-advantages.
It is very important to realise that “a poor design cannot be salvaged by good statistics”. A good study
design should be sound and scientific and planned very methodically. Therefore, an epidemiologist and statistician should be on board from the start of the study.
The epidemiologic study designs are broadly classified into “Observational” and “Experimental” study designs. Further classifications of various types of studies is outlined in Table 1.
A. OBSERVATIONAL STUDY DESIGNS
In an observational study design, the researcher or the investigator simply observes the individuals in the study and does not intervene in any way. They are further classified into “Descriptive” and “Analytical” studies.
DESCRIPTIVE STUDIES- These describe certain observations and could be in form of case reports or case series. Descriptive studies can be used
INDIAN JOURNAL OF MEDICAL SPECIALITIES 2012;3(2):191-194
Indian Journal of Medical Specialities, Vol. 3, No. 2, Jul - Dec 2012 191
for hypothesis generation and can also suggest associations. However, they cannot be used for testing the hypothesis.
ANALYTICAL STUDIES
Ecological studies- are usually undertaken to study the prevalence or incidence of diseases in populations or groups wherein the unit of study is population group rather than the individual [2]. For example, if we study the frequency of a characteristic (e.g. Alcohol intake) and some outcome of interest (e.g. Cirrhosis of Liver) occurring in the same geographic location (e.g. a city, state or a country). These studies can be used for generating hypotheses but not to draw causal conclusions because we do not have information as to whether people who consumed alcohol are the same people who developed cirrhosis of liver.
Cross-sectional study- In a cross-sectional study, data is collected at a single point in time or in a time frame. It is helpful to describe associations and determine prevalence. An example of a cross-sectional study is “Prevalence of Hypertension in Adults in a community” as shown in Figure 1.
The strengths of cross-sectional studies are that they can be performed quickly, are relatively inexpensive, provide the prevalence of a disease/risk factor and are useful to formulate a hypothesis. In spite of these salutary advantages, it has its
inherent weaknesses viz. cross-sectional studies cannot establish cause-effect relationship and cannot be used to test a hypothesis.
Case-Control Study- In a case-control study, one starts with people who have disease (cases) and then matches them with controls that do not have the disease. Then the researcher looks back and assesses the exposures in both the groups (cases as well as controls). A diagrammatic representation of a case control study is depicted in Figure 2.
The strengths of case-control studies are that they can be done in a short-period of time, are relatively inexpensive, are good for rare outcomes like cancer and can examine many exposures. Besides, they are useful to generate hypothesis and also are helpful in providing the odds ratio. The weak points include, that they cannot be used to determine the incidence, prevalence or the relative risk. Besides, they can only study one outcome and have a high susceptibility to bias.
CROSS-SECTIONAL STUDY
Sample of Population
Time Frame = Present
Obese Adults Prevalence of Hypertension
Non-obese Adults Prevalence of Hypertension
Figure 1- Depiction of a cross-sectional study
High E2 exposure
Low E2 exposure
High E2 exposure
Low E2 exposure
Patients with Uterine cancer (Cases)
Patients without Uterine cancer (Controls)
PAST PRESENT
Figure 2- Example of a Case-Control Study- Are those with uterine cancer more likely to have consumed Estrogen (E2)?
Table 1- Types of scientific study designs
A) OBSERVATIONAL - DESCRIPTIVE o Case Reports o Case Series
- ANALYTICAL o Ecological studies o Cross-sectional o Case-control o Cohort
B) EXPERIMENTAL o Randomised controlled clinical trial (RCCT) o Randomised controlled field trial (RCFT)
C) META ANALYSIS/SYSTEMATIC REVIEWS
Anita S Acharya and others
192 Indian Journal of Medical Specialities, Vol. 3, No. 2, Jul - Dec 2012
The third type of analytical observational studies commonly planned are the Cohort studies. Cohort studies are the most difficult ones; they begin with disease-free patients/subjects and classify them as exposed/unexposed. After, this the outcomes are recorded in both the groups and the outcomes are compared using relative risk.
Example of a Cohort Study is- “To determine the effects of smoking on lung Cancer mortality in a population”.
In a prospective cohort study, exposure may or may not have occurred at study entry, but outcome definitely has not occurred at time of study entry. The subjects are followed up for a fixed period of time or till the disease occurs. In a retrospective (historical) cohort study, both the disease and exposure have already occurred at the time of study.
The advantages of cohort studies are that they provide incidence data, help in establishing time sequence for causality, eliminate recall bias, can study rare exposures and allow for accurate measurement of exposure variables.
Cohort studies can measure multiple outcomes and can adjust for confounding variables and can also calculate relative risk. Despite the several advantages, there are weaknesses as well and these include- cohort studies are expensive, time-consuming, cannot study rare outcomes and require a relatively large sample size. Besides, exposure may change over a period of time, there is attrition of study population over time and diseases which have a long pre-clinical phase can be problematic to study as well.
B. EXPERIMENTAL STUDIES
Analytical experimental/interventional studies are studies in which an intervention is performed and hence, also are referred to as “clinical trials”. Clinical trials provide the “gold standard” for determining the strongest evidence for concluding causation. Various instruments which are employed during clinical trials include- - Randomisation - Blinding o Placebo-controlled or o Standard treatment
Randomisation literally means to toss a coin to decide the assignment of the patient to a study group. The most critical element in randomisation is the unpredictability of the next assignment which essentially ensures elimination of selection bias [3].
Blinding or masking involves many components. The subjects should not know to which group they are assigned to. This can be either done by using a “placebo” or where standard treatments are available they should be given. For example, to study a new drug against HIV/AIDS in a study group, standard treatment should be provided to the control group as it would be unethical to give a placebo to the control group.
When the study subjects are only blinded, it is termed as “Single Blind study” whereas when both the study subjects as well as data collectors are blinded, it is a “Double-blind” study.
There are two types of randomised trials: (1) Randomised Controlled Clinical Trial (RCCT) (2) Randomised Controlled Field Trial (RCFT)
Example of a Randomised Controlled Clinical Trial (RCCT) is depicted in Figure 3- To study the association between garlic consumption and cardiovascular disease prevention in a city.
Randomised Controlled Field Trial (RCFT): It is similar to an RCCT except that the intervention is preventive and not therapeutic. These are usually preventive trials in which the efficacy of a preventive
Study Population
Randomise to two groups
Control Group (Placebo)
Outcomes (CAD or no CAD)
Treatment group(Garlic pill)
Outcomes (CAD or no CAD)
Figure 3- Example of a Randomised Controlled Clinical Trial
Scientific study designs
Indian Journal of Medical Specialities, Vol. 3, No. 2, Jul - Dec 2012 193
intervention such as a new vaccine is tested in one study group and the other group receives a placebo or standard. As they are usually conducted in the community, the term used is Randomised Controlled Field Trial (RCFT).
The advantages and disadvantages of a RCCT and RCFT are similar. The strengths are that they are the best measure of a causal relationship, are the best design for controlling bias and can measure multiple outcomes. The weaknesses include the high cost, compliance issues and at times, ethical issues may be a problem. Sometimes, it may take a long time to obtain results. Another disadvantage of both is “External Validity” which is the ability to generalise the findings to other groups of population [1].
C. META ANALYSIS/SYSTEMATIC REVIEWS
It is analysis of multiple studies including statistical techniques for merging and contrasting results across studies [4]. The need for a meta-analysis or a systematic review arises when there are conflicting evidences available from different studies addressing the same research question. However a number of biases can occur like publication bias, aggregation bias, and bias in exclusion of studies. Detailed description is beyond the purview of this article.
Table 2 provides a tabulated summary of the characteristics of various study designs.
Table 2- Salient features of various study designs.
Key Points
• Choosing an appropriate study design to address a research question is very critical to obtain valid results.
• For hypothesis generation, observational, descriptive studies are generally used whereas for generation as well as hypothesis testing, observational, analytic studies like case-control, cohort studies are commonly employed.
• Randomised controlled trials are the “Gold Standard” for determining the strongest evidence for concluding causation.
The above treatise should help clarify the doubts of scientific personnel and help them in planning what study design they require to achieve their respective research questions.
Anita S Acharya and others
194 Indian Journal of Medical Specialities, Vol. 3, No. 2, Jul - Dec 2012
Symposium
Selecting a study design for researchPikee Saxena*, Anupam Prakash**, Anita S Acharya#, Aruna Nigam##
Departments of *Obs. & Gynae., **Medicine and #Community Medicine, Lady Hardinge Medical College & SSK Hospital, New Delhi-110001. ##Department of Obs. & Gynae., HIMSR & HAH Centenary hospital, New Delhi, India.Corresponding author: Dr. Pikee Saxena, Associate Professor, Department of Obs. & Gynae., Lady Hardinge Medical College & SSK Hospital, New Delhi-110001. India. Email- [email protected]: 11-06-2013 | Accepted: 26-06-2013 | Published Online: 07-07-2013This is an Open Access article distributed under the terms of the Creative Commons Attribution License (creativecommons.org/licenses/by/3.0)Conflict of interest: None declared | Source of funding: Nil | DOI: http://dx.doi.org/10.7713/ijms.2013.0033
Abstract
Study design forms the backbone of good research. It determines the methods and frequency of data collection, statistical tests to be applied and also helps in correct interpretation of data. Choosing a study design depends on several factors like research question, experience of the researcher, cost, time duration, latency and frequency of a disease. Study designs can be broadly categorised into observational studies where the investigator just observes the course that nature is taking without any manipulation; while experimental studies involve some intervention by the investigators in order to test the research hypothesis.
Observational studies could be descriptive or analytical. Descriptive studies help to generate a hypothesis like a case study or case series while analytical studies not only generate but also test hypothesis like a cross-sectional, case-control or cohort study. Experimental studies test the hypothesis and help to determine the cause of a disease. Each study design has its own merits and demerits which will be discussed in this article.
Epidemiology is a systematic study of the patterns, causes, and effects of disease conditions in a defined population. The primary goal of epidemiology is to identify causes of disease, investigate how outbreaks occur and explore ways to control and prevent diseases. Most epidemiological studies result in either formation of hypothesis or test the strength of association of this hypothesis in an unbiased manner. In simpler words, these study designs help to determine if the exposure is related to the disease.
What is a study design?
Design is the pattern, scheme or plan to collect evidence. Study design is a tool on which the credibility of research findings depends. Epidemiological studies are aimed at examining relationships between exposures like alcohol or smoking, biological agents, stress, or chemicals
to disease outcome. The identification of causal relationships between these exposures and outcomes is an important aspect of epidemiology. They also help to determine whether the results obtained are scientifically sound and unbiased.
Study design depends on a multitude of factors [1] like research question and goal, skills of the researcher, availability of funds, time duration during which a study is to be conducted, existing knowledge about the research question/disease, duration of latency of a disease and on whether it is a rare or common disease.
On the basis of any intervention done by the investigator, study designs can be divided into two broad categories (Table 1) [2,3]-
1. Observational studies are those where the investigator studies people or exposure as they exist or occur in nature. These do not involve any intervention or experiment by the investigator.
334 Indian Journal of Medical Specialities, Vol. 4, No. 2, July - Dec 2013
INDIAN JOURNAL OF MEDICAL SPECIALITIES 2013;4(2):334-339
It is of two types-
A. Descriptive studies are used to formulate hypothesis or association. They describe the frequency or characteristics of an outcome. They commonly answer the questions- who, when, what and where of a disease occurrence [2]. Examples include case-studies and cross-sectional studies.
B. Analytical studies are used to test hypotheses or the relationship between exposure and outcome. They commonly answer the question how a disease occurs? Examples are cross-sectional, case-control, cohort studies.
2. Experimental studies involve manipulation or intervention by the investigators in order to test the hypothesis through experiment. Examples are the randomised control trial (often used for new drug or comparing drug therapies), field trial (conducted on those at a high risk of contracting a disease) and community trial (research on social diseases).
In order to identify the aetiologic factors or exposure responsible for an outcome, each step in the epidemiologic framework provides new and important information. Depending on the increasing knowledge of disease or exposure and generating evidence, they can be classified in to a hierarchy as-
Cross-sectional studies –Help to develop hypothesis↓
Case Control study– Investigate relationship of exposure to outcome
↓Cohort study- Defines strength of association between exposure and outcome
↓Control Trial-Tests associations experimentally
Observation tools for data collection in a study:
• Selected units could be either individuals or a defined group of individuals/population.
• Based on whether single or multiple observations are to be made studies could be categorised as:
a. Cross-sectional study where only single or one set of observations are made, at a certain point in time.
b. Longitudinal study where at least two or more sets of observations are collected at different points of time.
• Based on data collection timing:
a. Prospective study looks forward in time and involves follow up. Example: In a prospective cohort, study is started with a cohort of disease free individuals who are followed up in time to observe how many of them develop a disease.
b. Retrospective study looks backwards in time. Example: Case control study, retrospective cohort study.
• Based on data collection methods:
a. Primary data is where the investigator is collecting the data. Sources include medical examinations, interviews, etc. Advantages of collection of primary data are that there is less measurement error and it suits objectives of the study better. The disadvantage is that it is costly or it may not be feasible.
b. Secondary data is where the data is collected by others and has not been collected for the purpose of the study. Sources include census data, vital statistics.
Descriptive Studies
1. Case Reports or case series are detailed presentation of a single or a group of cases having unusual symptom or response. In many ways they are “sentinel events” which may lead to testable
Table 1- An overview of the various types of study designs
Observational Study Experimental Study
A. Descriptive study Randomised Control Trial Case study Field Trial Case Series Community trial
B. Analytical study Cross-sectional study Case-control study Cohort study
Indian Journal of Medical Specialities, Vol. 4, No. 2, July - Dec 2013 335
Research study designs
hypothesis. The case is described along with the help of radiological, pathological, biochemical or/and haematological investigations. Case studies generally report a new or unique finding e.g. previous undescribed disease, unexpected link between diseases, unexpected new therapeutic effect or unexpected adverse events. A large number of case reports are indexed on Medline (www.pubmed.com) which provide a useful reminder about rare conditions, diagnoses, etc.
Advantages of case series are that they are useful for hypothesis generation and they provide information about rare diseases with few established risk factors. However, they cannot study cause and effect relationships and cannot assess disease frequency.
Examples of case report-
Generalised lymphadenopathy- an unusual aetiology;
A case report of a live, intrahepatic pregnancy;
Sudden death after administration of oral azithromycin.
Analytical Studies
1. Cross-sectional study represents a “Snapshot” of health problems of a population at a particular time. Both the exposure and disease are measured at the same time. These studies can be done quickly, are less expensive and do not require any follow up. They describe associations & measure point prevalence i.e. they determine the total number of persons suffering from a disease including both new and old cases at a particular time. They are useful for estimating population burden, health planning and priority setting of health problems. They are not useful for studying rare diseases. As compared to longitudinal studies there is no risk of individuals being lost to follow up.
Cross-sectional study calculates odds ratio (OR) which is a measure of association between an exposure and an outcome [4].
OR=1 Implies that exposure does not affect odds of outcome.OR>1 Implies that exposure is risky and is associated
with higher odds of outcome. OR<1 Implies that exposure is protective and is associated with lower odds of outcome.
The limitations of a cross-sectional study are that they have comparatively weaker observational design as they measure prevalence and not the incidence (new cases) of a disease. Another problem is that they consider only survivors and do not take into account the persons who have died due to the disease. They cannot determine the temporal relationship between the exposure and the disease and therefore cause-effect relationship cannot be established.
Examples of Cross-Sectional Studies-
• Is there any association between oat consumption and coronary artery disease (CAD)? (Figure 1)
• To determine height and weight pattern of students of class VIII.
• To assess prevalence of infertility amongst women attending Gynaecological OPD
2. Case control study is a retrospective study where the starting point is the identification of ‘cases’ with disease and suitable ‘controls’ without that disease. Both the cases and controls are selected ideally from the same population. Cases and controls are then compared to assess whether there were any differences in their past exposure. They are useful for generating hypothesis. These studies are most suitable for rare diseases or diseases with long latent period. The statistic generated to measure
Figure 1- Diagrammatic depiction of a cross-sectional study design
336 Indian Journal of Medical Specialities, Vol. 4, No. 2, July - Dec 2013
Pikee Saxena and others
association is the odds ratio, which is the ratio of the odds of exposure in the cases to the odds of exposure in the controls. Case control studies provide low cost answers to health questions. They are frequently used for investigating outbreak of a disease. The other advantages of these studies are that they are less expensive, consume less time. They can examine many exposures at one time.
The limitations of these studies are that they cannot measure incidence, prevalence or relative risk of exposure. These studies can only evaluate one outcome at a time. They are liable to recall & selection bias.
Examples of Case-Control Studies-
• Are individuals suffering from coronary artery disease less likely to have consumed oats as compared to controls? (Figure 2)
• To investigate the role of diet in breast cancer.• To assess whether hepatitis B infection played a
role in the aetiology of primary hepatocellular cancer?
3. Cohort Studies include a group of people who have something in common which may be a similar characteristic or exposure e.g. all people who attended a cricket match or all persons in a hospital who died due to a heart attack. This selected population is called a cohort. The study is initiated with disease-free patients who are followed up prospectively or retrospectively in time. The patients are then classified as exposed or unexposed. Data is collected at multiple points in time to identify new cases or incidence of a disease. These studies calculate relative risk which compares outcome in the exposed and unexposed group. Cohort studies are used to establish causation of a disease or to evaluate the outcome or impact of treatment. They are also useful for defining the spectrum of a disease.
In prospective cohort studies, exposure factors are identified at the beginning of a study and a defined population is followed into the future. In retrospective or historical cohort studies, past medical records for the defined population are used to identify exposure factors. In a retrospective (historical) cohort study, both the disease and exposure have already occurred at the time of study.Cohort studies provide incidence data and can
establish time sequence for causality of a disease outcome. They eliminate recall bias as the individuals are being observed directly and can measure multiple outcomes. Confounders can also be controlled more easily. They calculate relative risk (RR) or risk ratio which determines whether a disease is associated with exposure and also determines the strength of association. The RR is a more powerful effect measure than the OR, as the OR is just an estimation of the RR, since true incidence cannot be calculated in a case-control study where subjects are selected based on disease status. As with the OR, a RR=1 shows that there is no association between the exposure and disease, RR > 1 shows risky or positive association while RR <1 indicates negative or protective association.
Cohort studies are expensive and time consuming as the individuals are followed over a long period of time. They are not suitable for studying rare diseases or diseases with a long latent period and cannot prove causation. Other limitations are that due to long study duration, exposure may change over time and individuals may be lost during follow up.
Figure 2- Diagrammatic depiction of Case-control study design
Figure 3- Diagrammatic depiction of a prospective cohort study
Study population
Case control design
Oat eaters
Oat eaters
Oat non eaters
Oat non eaters
CAD (disease)
Controls (no disease)
}}
}
Study begins here
Time
Study population
free of disease
Cohortdesign Oat eaters
Oat non eaters
CAD
CAD
(no disease)
(no disease)
}}
}
timeStudy begins here
futurepresent
Indian Journal of Medical Specialities, Vol. 4, No. 2, July - Dec 2013 337
Research study designs
Examples of Cohort Studies-
• To study the effects of oat consumption as compared to oat non-consumption on CAD in a population? (Figutre 3)
• To investigate the association between hepatitis B virus infection and the development of primary hepatocellular carcinoma?
• Whether family income affects educational achievement?
4. Randomised control trial (RCT) is an experimental study which is considered to be the “Gold Standard” in determining the relationship between exposure and outcome and generates Level 1 evidence [5]. It is used in clinical trials to evaluate whether an intervention or treatment works or to establish the efficacy and safety of medical technologies and health services.
Special characteristics of RCT are that the subjects are randomly assigned to “treatment or study” and “control” groups. Both these groups are similar to each other except for the intervention being tested. Randomisation helps to avoid the bias in choice of patients for treatment by a physician. This increases the probability that differences between the groups can be attributed to the treatment under study. After randomisation, allocation concealment is done for protecting randomisation process so that the treatment to be allocated is not revealed to the investigators before the patient is entered into the study, in order to avoid selection and confounder bias. Methods used for allocation concealment are sequentially numbered, opaque, sealed envelopes; pharmacy controlled or centrally controlled randomisation. This is followed by blinding procedures which are done to prevent participants, care-givers, or outcome assessors from knowing which intervention was received to
eliminate bias in outcome measurement. CONSORT 2010 statement recommends that it should be specified if participants, care providers or analysers were blinded rather than just mentioning “single-blind,” “double-blind,” and “triple-blind’ depending on whether one or two or all three parties were blinded.
An RCT is the best measure of causal relationship and is most effective design for controlling bias. Multiple outcomes can be evaluated in a RCT. However, RCTs are energy and time consuming. They are relatively costly and may not be feasible for all interventions or settings. They are mostly able to address one question at a time as studying multiple variables at one time may lead to errors. There is also an ethical concern of not giving intervention to the control group patients. RCTs are not practical for studying rare events. As RCTs are conducted in controlled conditions there is a limitation of external validity or generalisation of the results [6].
Examples of RCT-
• Does oat consumption reduce CAD as compared to those who are not exposed to oat intake? (Figure 4)
• The Women’s Health Initiative is a randomised trial in the USA to determine whether a sustained low-fat diet will reduce the incidence of breast cancer.
• Before and after measurements in a clinical trial for a new therapy, laboratory experiments, field trials.
Conclusions
There may be an overlap in the conceptual basis of quantitative study designs. If the cross-sectional study is repeated and the same sample is studied for a second time, the original cross-sectional study will become a cohort study. If, during a cohort study, the investigator evaluates an intervention, it becomes a trial. Cohort study also may change to case-control studies, using cases evolving during the course of the study (nested case control study). Cases of a case-series in a fixed population, may initiate a case-control study or a trial. It should be remembered that every epidemiological study may not fit neatly into one of the basic designs.
Figure 4- Experimental design of an RCT
338 Indian Journal of Medical Specialities, Vol. 4, No. 2, July - Dec 2013
Pikee Saxena and others
References
1. Acharya AS, Prakash A, Nigam A, Saxena P. Scientific study designs for research: an overview. Indian J Med Spec 2012;3:191-4.
2. Das R, Das PN. Biomedical Research Methodology including Biostatistical Applications. Jaypee Brothers Medical Publishers, New Delhi 1st edition, 2011; pp.19.
3. Rothman KJ, Greenland S, Lash TL. Modern Epidemiology. Types of epidemiologic studies: clinical trials. Lippincott Williams & Wilkins, Philadelphia, 3rd edition, 2008; pp. 89–92.
4. Prakash J, Nigam A, Saxena P, Acharya AS. Statistical tests of significance: an overview of the basic concepts. Indian J Med Spec 2012;3:203-6.
5. http://www.cebm.net/index.aspx?o=1025. Centre for Evidence-based Medicine. Levels of evidence. Last updated 12 July, 2012. Accessed on August 10, 2012.
6. Saxena P, Prakash A, Acharaya AS, Nigam A. How to design and conduct a randomised controlled trial? Indian J Med Spec 2012;3:198-202.
Key Points
• Study design is the basic framework of research and should be carefully selected before initiation of the study as it is the most essential tool on which the credibility of the study depends.
• Descriptive studies like case study and case series help to generate hypothesis.
• Cross-sectional study is a snapshot of a health outcome at a particular point in time. It helps to calculate point prevalence of a disease. Odds ratio can be calculated.
• Case-control study is a retrospective study which measures prevalence of risk factors in diseased and non-diseased individuals. It is useful for rare diseases. Odds ratio can be calculated.
• Cohort study could be prospective or retrospective where a group of individuals with a common characteristic are followed either into future or backwards into time. They test the hypothesis and calculate incidence, prevalence and relative risk. However, they cannot prove causality.
• Experimental studies include randomised
Table 2- A comparison of various study designs
Cross sectional Case control Cohort RCT
Comparative Cost Least Less More Most
Duration of study Relatively Less Relatively Most time- Relatively More consuming More
Different study designs answer different clinical questions. Each study design has its own characteristic strengths and limitations. The various study designs have been compared in Table 2. No single study is perfect and clinical decisions should be based on consistent findings from several studies that use similar design and methodology.
control trials which test the hypothesis experimentally and prove causality. They include special measures like randomisation, blinding, allocation concealment and are carried out under controlled conditions. They are the “Gold Standard” in determining the relationship between exposure and outcome and generate Level 1 evidence.
Indian Journal of Medical Specialities, Vol. 4, No. 2, July - Dec 2013 339