Sample size calculator for cluster randomized trials

Health Services Research Unit, University of Aberdeen, Polwarth Building Foresterhill, Aberdeen, Scotland AB25 2ZD UK.
Computers in Biology and Medicine (Impact Factor: 1.24). 04/2004; 34(2):113-25. DOI: 10.1016/S0010-4825(03)00039-8
Source: PubMed


Cluster randomized trials, where individuals are randomized in groups are increasingly being used in healthcare evaluation. The adoption of a clustered design has implications for design, conduct and analysis of studies. In particular, standard sample sizes have to be inflated for cluster designs, as outcomes for individuals within clusters may be correlated; inflation can be achieved either by increasing the cluster size or by increasing the number of clusters in the study. A sample size calculator is presented for calculating appropriate sample sizes for cluster trials, whilst allowing the implications of both methods of inflation to be considered.

196 Reads
  • Source
    • "These interventions were selected because they are the most effective based on their effect sizes [3] [14] [17], yet the least used [9] [20] [21]. Based on a control group mean number of interventions = 0.1, SD = 0.3 [20], D = 0.5, intracluster correlation coefficient = 0.6 [2], a = 0.05, and b = 0.8, 136 participants were required, equivalent to 14 classes, with up to 10 participants/class. The sample size was doubled to 28 classes to account for fewer individuals/class and losses to follow-up. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Educating parents about ways to minimize pain during routine infant vaccine injections at the point of care may positively impact on pain management practices. The objective of this cluster randomized trial was to determine the impact of educating parents about pain in outpatient pediatric clinics on their use of pain treatments during routine infant vaccinations. Four hospital-based pediatric clinics were randomized to intervention or control groups. Parents of 2- to 4-month-old infants attending the intervention clinics reviewed a pamphlet and a video about vaccination pain management on the day of vaccination, whereas those in the control clinics did not. Parent use of specific pain treatments (breastfeeding, sugar water, topical anesthetics, and/or holding of infants) on the education day and at subsequent routine vaccinations 2 months later was the primary outcome. Altogether, 160 parent-infant dyads (80 per group) participated between November 2012 and February 2014; follow-up data were available for 126 (79%). Demographics did not differ between groups (P > 0.05). On the education day and at follow-up vaccinations, use of pain interventions during vaccinations was higher in the intervention group (80% vs 26% and 68% vs 32%, respectively; P < 0.001 for both analyses). Educating parents about pain management in a hospital outpatient setting leads to higher use of pain interventions during routine infant vaccinations.
    Pain 01/2015; 156(1):185-91. DOI:10.1016/j.pain.0000000000000021 · 5.21 Impact Factor
  • Source
    • "An absolute increase in FOBT screening rates of 15% from the most current FOBT completion rate know at the time of the study (42% in Manitoba) was considered clinically significant [15]. Sample size was determined by a biostatistician using PASS (Power and Sample Size, 2002) software [19] and a computer simulated ICC value of 0.6 [20]; a very conservative estimate of the minimum detectable effect size given the proposed sample size. Based on logistic regression, a cluster size of 41 clusters each enrolling 30 to 35 patients (1230 observations) will achieve 90% power at a 0.05 significance level to detect a change from the baseline value of 0.400 to 0.550 and corresponds to an odds ratio of 1.833. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Evaluation of the effectiveness of a patient decision aid (nurse-managed telephone support line and/or colorectal cancer screening website), distributed to patients by their family physician, in improving fecal occult blood test (FOBT) colorectal cancer screening rates. A pragmatic, two arm, cluster randomized controlled trial in Winnipeg, Manitoba, Canada (39 medical clinic clusters; 79 fee-for-service family physicians; 2,395 average risk patients). All physicians followed their standard clinical screening practice. Intervention group physicians provided a fridge magnet to patients that facilitated patient decision aid access. Primary endpoint was FOBT screening rate within four months. Multi-level logistic regression to determine effect of cluster, physician, and patient level factors on patient FOBT completion rate. ICC determined. Family physicians were randomized to control (n = 39) and intervention (n = 40) groups. Compared to controls (56.9%; n = 663/1165), patients receiving the intervention had a higher FOBT completion rate (66.6%; n = 805/1209; OR of 1.47; 95% confidence interval 1.06 to 2.03; p < 0.02). Patient aid utilization was low (1.1%; 13/1,221) and neither internet nor telephone access affected screening rates for the intervention group. FOBT screening rates differed among clinics and physicians (p < 0.0001). Patients whose physician promoted the FOBT were more likely to complete it (65%; n = 1140/1755) compared to those whose physician did not (51.1%; n = 242/470; p < 0.0001; OR of 1.54 and 95% CI of 1.23 to 1.92). Patients reporting they had done an FOBT in the past were more likely to complete the test (70.6%; n = 1141/1616; p < 0.0001; 95% CI 2.51 to 3.73) than those who had not (43%; n = 303/705). Patients 50–59 years old had lower screening rates compared to those over 60 (p < 0.0001). 75% of patients completing the test did so in 34 days. Despite minimal use of the patient aid, intervention group patients were more likely to complete the FOBT. Powerful strategies to increase colorectal cancer screening rates include a recommendation to do the test from the family physician and focusing efforts on patients age 50–59 years to ensure they complete their first FOBT. Trial registration Trial registration number: identifier NCT01026753.
    BMC Cancer 04/2014; 14(1):263. DOI:10.1186/1471-2407-14-263 · 3.36 Impact Factor
  • Source
    • "A standard sample size formula was used to calculate the initial unadjusted sample size requirements, followed by appropriate adjustment for clustering by general practice according to Campbell et al.[23], with expected small clustering effect (ICC = 0.05). Previous reported ICCs vary from 0.03 for smoking advice [23] to 0.05 for cluster randomized trials in primary care (data from Trial of Older People in the Community) [24,25]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Chronic obstructive pulmonary disease (COPD) remains a major health problem, strongly related to smoking. Despite the publication of practice guidelines on prevention and treatment, not all patients with the disease receive the recommended healthcare, particularly with regard to smoking cessation advice where applicable. We have developed a tailored implementation strategy for enhancing general practitioners' adherence to the disease management guidelines. The primary aim of the study is to evaluate the effects of this tailored implementation intervention on general practitioners' adherence to guidelines.Methods/design: A pragmatic two-arm cluster randomized trial has been planned to compare care following the implementation of tailored interventions of four recommendations in COPD patients against usual care. The study will involve 18 general practices (9 in the intervention group and 9 in the control group) in Poland, each with at least 80 identified (at the baseline) patients with diagnosed COPD. The nine control practices will provide usual care without any interventions. Tailored interventions to implement four recommendations will be delivered in the remaining nine practices. At follow-up after nine months, data will be collected for all 18 general practices. The primary outcome measure is physicians' adherence to all four recommendations: brief anti-smoking advice, dyspnea assessment, care checklist utilization and demonstration to patients of correct inhaler use. This measurement will be based on data extracted from identified patients' records. Additionally, we will survey and interview patients with chronic obstructive pulmonary disease about the process of care. The results of this trial will be directly applicable to primary care in Poland and add to the growing body of evidence on interventions to improve chronic illness care.Trial registration: This trial has been registered with Clinical Trials Protocol Registration System. Trial number: NCT01893476.
    Trials 04/2014; 15(1):109. DOI:10.1186/1745-6215-15-109 · 1.73 Impact Factor
Show more