The intraclass correlation coefficient (ICC) measures the similarity of observations within clusters and affects the sample size needed in cluster sampling. A high ICC indicates that individuals within clusters are more alike, necessitating a larger sample due to less independent information per individual. In the study, ICC is used in the formula DEFF = 1 + (m - 1) × ICC. This affects the study design by determining the extent to which clustering inflates the required sample size. Here, ICC contributes to setting DEFF at 1.3, reflecting moderate clustering effects .

The calculated values 'n0', 'n_deff', and 'n_final' form a structured approach to ensuring the study is statistically robust. 'n0' is the initial sample size based solely on precision requirements. 'n_deff' accounts for clustering by inflating 'n0' using DEFF, addressing the design's complexity. 'n_final' adjusts ‘n_deff’ for non-response rates, determining the total number needed to reach valid sample conclusions. These values ensure the study effectively addresses sampling challenges while achieving reliable and representative results .

R programming enhances the efficiency and accuracy of sample size calculations in public health studies through reproducible scripts and functions. In the article, functions such as z_alpha_2 and ss_single_prop are defined for calculating Z-scores and initial sample sizes, respectively. Adjustments for design effect and non-response are handled by wrapper functions like inflate_deff and inflate_nr, which automate calculations and ensure consistency across trials. Such scripts facilitate simulations and sensitivity analyses, allowing researchers to adjust parameters flexibly and observe varying outcomes efficiently .

Non-response inflation accounts for potential non-responses in the sample, ensuring that the final sample size meets the study's needs. It is calculated by dividing the inflated sample size by 1 minus the non-response rate (nfinal = n0 × DEFF / (1 - r)). In this study, a non-response rate (r) of 15% is used, ensuring that the final sample size accommodates respondents who might not participate, thus increasing the reliability of the study findings .

Setting different margin of error levels helps evaluate the trade-off between precision and resource requirements in a study. Smaller margins of error result in higher sample sizes, providing more precise estimates but requiring more extensive resources in terms of time and cost. In the study on glaucoma management, this approach allows researchers to see how changes in margin of error affect the number of respondents needed and budget allocations. This analysis helps in strategic decision-making regarding the study's feasibility and reliability under varying logistic constraints .

The choice of clinic attendee fraction (π) directly influences the number of total attendees that need to be approached to achieve the desired sample size. A lower π indicates that a smaller proportion of attendees meet the study inclusion criteria, necessitating a larger number of initial contacts. In this study, with π set at 0.0936, a significant number of attendees must be approached to ensure an adequate sample size of glaucoma or glaucoma-suspect individuals. This adjustment highlights the logistic challenge of reaching enough eligible participants, impacting resource allocation and study timelines .

The design effect (DEFF) quantifies the inflation in sample size necessary when using cluster sampling compared to simple random sampling. In the study assessing knowledge, attitude, and practice (KAP) toward glaucoma management, DEFF is used to adjust for clustering effects. It is computed using the formula DEFF = 1 + (m - 1) × ICC, where m is the average cluster size and ICC is the intraclass correlation coefficient. In this study, DEFF was set at 1.3, indicating that the required sample size is 30% larger due to clustering effects .

The margin of error (d) directly affects the calculated initial sample size (n0). A smaller d requires a larger n0 to ensure a more precise estimate of the population proportion. In this study, different margins (0.05, 0.04, 0.03, 0.025, 0.02) were considered, resulting in varier n0 values, which then affected the final number of clinic attendees to approach. For example, at d = 0.05, n0 is 130, leading to 2,137 attendees, while at d = 0.02, n0 increases to 815, leading to 13,334 attendees .

A change in confidence level (α) affects the Z-score used in the sample size calculation formula, directly influencing the necessary sample size. A higher confidence level requires a larger Z-score, increasing the calculated sample size to ensure the results reflect greater certainty. In the study, the confidence level α is 0.05, corresponding to a Z-score of 1.96. Adjustments to α would alter this value, consequently recalibrating n0 to maintain precision and confidence in the study outcomes .

The methodological steps for using the single proportion formula in sample size estimation involve several calculations: First, determine the initial sample size (n0) using the formula n0 = Z^2(1−α/2) p(1 − p) / d^2, where Z is the Z-score for the desired confidence level, p is the estimated proportion, and d is the margin of error. Then, account for cluster sampling using the design effect (DEFF = 1 + (m -1) × ICC). Finally, adjust for non-response rate with nfinal = n0 × DEFF / (1 - r). These steps ensure a robust estimate accommodating various study design elements .