State the setting for your study? Suggest two different ways to recruit your study population. Discuss the benefits and limitations of each approach and decide (with justification) which one to implement in your study?
This research was conducted at a tertiary care hospital in Karachi, the capital of Sindh, in Pakistani. The study involved the ENT Department of Abbasi Shaheed Hospital and Karachi Medical & Dental College in collaboration with the microbiology department from October 2010 to January 2015 on 221 patients (Sajjad et al, 2020). There are two primary ways of recruiting a study population, probabilistic and non-probabilistic sampling.
Probabilistic sampling- this approach of recruiting a sample population emphasizes that all items of the total population have a nonzero chance of being recruited into the study. If all participants have an equal probability of being selected for a survey. Then an equiprobalilistic sampling is suitable, and the chances of a participant being recruited into the survey can be expressed by the principle
PROBABILISTIC SAMPLING AND NON PROBABILISTIC SAMPLING
P=1/N.
Where P represents the probability of being selected to the survey, and N represents the size of the target population. Some of the main kinds of probabilistic sampling include;
Simple random sampling- where individuals taking part in a survey are randomly selected from a full list of participants.
Systematic random sampling- this is where participants are recruited to a survey from fixed interims previously identified from a categorized list of participants.
Stratified sampling- this is where the target population is categorized into distinct strata, then samples are recruited to a survey from each category.
Cluster sampling- this is a type of sampling where groups such as schools, organizations, and firms are samples.
Multi-stage sampling- this is a type of sampling that involves a combination of several strategies in recruiting sample units, such as simple random sampling in the first stage and stratified sampling in the next step.
Non-probabilistic sampling- in this kind of sampling, the probability of recruiting sample units from the target population is null. This form does not render a representative sample; thus, the outcomes of the study cannot be generalizable to the target population. Some of the types under this method of sampling are;
Convenience sampling- this is a sampling process where the participants are successively recruited in order of appearance according to their suitable convenience.
Purposive sampling- this is a sampling process when a varied sample is needed, or the judgment of professionals in a specific field is a subject of concern.
Quota sampling- this is a sampling technique where the population is first categorized by attributes such as gender or age. Sample units are then recruited to a survey to complete each proportion.
Snowball sampling- this is a technique where the researcher recruits participants, and then the participants later indicate additional potential associates to take part in the survey.
The main advantage of non-probabilistic technique as compared with probabilistic sampling is that it is less costly and time-saving. Besides, it is simple to utilize, especially when it seems hard to undertake a probabilistic inspection (especially in circumstances of a low target population). Nevertheless, the non-probabilistic technique becomes ineffective in determining how well the researcher is speaking to the target population. Therefore, the results cannot be generalized to the target population. On the other hand, probabilistic has a higher level of reliability in its research findings.
Besides, it has a lot of possibilities of making inferences about the population under study, unlike non-probabilistic sampling. Nevertheless, it is more complex and time-consuming than non-probabilistic sampling. From the above assumptions, non-probabilistic sampling comprises of much more substantial drawbacks. Since this study seeks an outcome that can be used to generalize the target population, I could recommend a multi-stage sampling technique. In this regard, purposive sampling would be utilized to sample patients who have ever received a clinical diagnosis of nasal polyposis with or without fungal infection and those who have not. Random sampling would also be utilized, and lastly, snowballing technique.
Define your primary outcome (or target condition) and explain why you have selected it. Suggest two different ways in which it could be measured/assessed. Discuss the advantages and disadvantages of each approach and decide (with justification) which one should be implemented in your study.
The primary outcome measure of this study is the prevalence of fungal infection in nasal polyposis among patients treated at the facility under investigation. I identified this outcome on the basis that it will reduce the false-negative inaccuracy by offering the basis for estimating the sample size that is adequate for a sufficiently powered research. Secondly, this selection will lessen the false-positive error arising from the statistical analysis of several outcomes. There are several ways in which the primary outcome can be assessed in a study, first, through regression analysis. The primary outcome is what the researcher considers most prominent among the several results that are to be analyzed in the research. If a study is conducted for a primary outcome measure, but instead, a secondary outcome is selected to be stated for statistical significance, then the study might yield false-negative conclusions.
Therefore, regression allows the researcher to examine the relationship between variables under research to generate the most accurate results. The second way that could measure the primary outcome is through the structural equation model (SEM) after the collection of data. SEM consists of a multivariate statistical analysis method that analyses structural relationships. It combines both factor analyses and multiple regression to analyze the relationship between variables in a study. As compared with regression analysis, SEM is better because it estimates the varied and interconnected dependence of variables in a single assessment, unlike regression that requires simultaneous analysis of the variables. Besides with SEM measurement error is not quantified in a residual error expression. Therefore, I would implement an SEM measurement technique to determine the variability and validity of model measures.
State what your unexposed group/comparator group/reference test is. Describe how and when it would be assessed/implemented. Justify your answers.
The unexposed group in this study are the patients who had undergone a clinical diagnosis of nasal polyposis without fungal infections based on nasoendoscopic examinations. The comparator group allows for comparison, offering an evaluation of the baseline or expected prevalence of fungal infections in nasal polyposis among patients. The comparator group is assessed after the convincing results of the analytic epidemiology have been analyzed. The researcher develops a hypothesis regarding the patterns and about the prevalence of fungal infection in nasal polyposis. The results of the study are then compared with this comparison group. If the prevalence rate is practically dissimilar in the exposed group as matched with the unexposed group, the exposure is then established to be associated with infection.
Using the existing literature, provide the relevant information that you would need to estimate the number of participants in a study that was sufficiently powered to detect a clinically relevant effect of an intervention/exposure on the outcome of interest, or to obtain a sufficiently precise estimate of diagnostic accuracy. Carry out a sample size calculation. Clearly state and defend your choice of the required sample size. These?
Every research needs to be properly planned, starting from the objective of a trial to statistical methods and conclusions; all these processes should be adequately planned to ensure valid results. Engaging many participants is expensive and exposes many participants to the procedure. In contrast, fewer participants will result in a statistically inconclusive study and might fail the whole protocol. Therefore, sample size estimation is a crucial aspect of research. According to Lutz (1982), there are some basic principles to be followed in sample estimations in a clinical setup. They include;
- The level of significance is common takes as 0.05, and in this regard, the sample size should increase while the level of significance decreases.
- The level of power should be >= 80%, whereas the sample size should increase as power increases. If the power is high, then the chances of missing the real impact are low. There should be a clinically meaningful distinction
Evidence from past research estimates about 20% prevalence of fungal infections in the population being surveyed. It is assumed that the level of significance is 5% and a 10% margin of error.
Therefore the sample size equated as N= (Zα/2)2 P (1-P)*1 / E2
3.8416*0.16/0.0004= 1536
Thus a sample size of 1536 is needed to participate in a study to approximate the prevalence of fungal infection. (E is termed as the margin of error and is provided as 0.02).
