See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/336762026

How to Calculate Sample Size for Observational and Experimental Nursing

Research Studies?

Article in National Journal of Physiology Pharmacy and Pharmacology · November 2019

DOI: 10.5455/njppp.2020.10.0930717102019

CITATIONS READS

20 3,434

4 authors:

Suresh K Sharma Shiv Mudgal

All India Institute of Medical Sciences Jodhpur Pacific Medical University
143 PUBLICATIONS 371 CITATIONS 37 PUBLICATIONS 67 CITATIONS

SEE PROFILE SEE PROFILE

Kalpana Thakur Rakhi Gaur

All India Institute of Medical Sciences Rishikesh Pacific Medical University, Udaipur
18 PUBLICATIONS 27 CITATIONS 16 PUBLICATIONS 37 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Burden of Dialysis among Caregivers at Tertiary care Hospital. View project

MSc Psychiatric Nursing Thesis Research View project

All content following this page was uploaded by Suresh K Sharma on 18 February 2020.

The user has requested enhancement of the downloaded file.

 National Journal of Physiology, Pharmacy and Pharmacology

REVIEW ARTICLE
How to calculate sample size for observational and experimental nursing
research studies?

Suresh Kumar Sharma, Shiv Kumar Mudgal, Kalpana Thakur, Rakhi Gaur

Department of Nursing, College of Nursing, All India Institute of Medical Sciences, Rishikesh, Uttarakhand, India
Correspondence to: Shiv Kumar Mudgal, E-mail: peehupari05@gmail.com

Received: September 08, 2019; Accepted: October 17, 2019

ABSTRACT

Sample size estimation for a research study is the most crucial part of the research process because it helps to produce
reliable results which improve generalizability of study results. A researcher must have understanding about significance
level, effect size, study’s power, and effect size; margin of error and ratio in event among population and design effect to
use sample size calculation formulas efficiently. There are different formulas of sample size calculation for different types
of variables measured in distinct study designs, namely descriptive, epidemiological, comparative, and interventional
research studies which are covered in this article. Review authors searched online and grey literature related to sample size
and read extensively. There were two authors who extracted and complied information related to topic.

KEY WORDS: Sample Size Calculation; Observational Studies; Experimental Studies; Case–Control and Cohort Studies

INTRODUCTION are available for sample size calculation when investigator

is dealing with different types of parameters and variables in
Sample size estimation is the most crucial methodological different study designs. When investigators have understanding
part of a research study for precisely drawing inferences about of basic statistical conceptions for sample size calculation, they
population or generalization of study results. A too large sample may also use available software for sample size estimation.
is merely a wastage of resources and time and on the other
hand, too small size sample fails to produce a conclusive and Review authors extensively searched, reviewed, and extracted
reliable results.[1] Therefore, it is essential for an investigator to information about basic statistical conceptions of sample size
estimate optimum sample size to produce reliable results, which calculation and provides an easy and practical approach of
can serve as strong foundation for evidence-based practices. It sample size calculation, while investigator is dealing with
has been observed that a large number of published researches different types of parameters and variables in observational
lacks clarity about the sample size estimation as well as indicate studies, randomized controlled trials, and diagnostic studies.
less power due to suboptimal sample size.[2-7] Sample size
estimation is also essential to know the feasibility of study in
terms of required cost and time. There is no magic solution for BASIC STATISTICAL CONCEPTIONS OF
sample size estimation; however, different statistical formulas SAMPLE SIZE CALCULATION

Access this article online A researcher needs to have the following information in hand
Website: www.njppp.com Quick Response code for the estimating sample size for a particular study: [8-10]

Null Hypothesis and Alternative Hypothesis

DOI: 10.5455/njppp.2020.10.0930717102019
Experimental and correlation observational studies are used
null hypothesis (indicate no observed difference), which are

National Journal of Physiology, Pharmacy and Pharmacology Online 2020. © 2020 Shiv Kumar Mudgal, et al. This is an Open Access article distributed under the terms of the Creative
Commons Attribution 4.0 International License (http://creative commons.org/licenses/by/4.0/), allowing third parties to copy and redistribute the material in any medium or format and to
remix, transform, and build upon the material for any purpose, even commercially, provided the original work is properly cited and states its license.

1 National Journal of Physiology, Pharmacy and Pharmacology 2020 | Vol 10 | Issue 01

Sharma et al. Calculation of sample size in nursing research

set to be rejected and always used in conjunction with an while calculating for sample size. It is usually estimated
alternative hypothesis (significant difference). Sometimes, it is from previous literature, including observational cohorts.
not possible that the null hypothesis can be rejected, but it does For example, studying the association of alcohol and liver
not mean that it is true, and it just states that we are unable to disease, the prevalence rate for liver disease in studying
produce enough evidence to reject stated null hypothesis. Thus, population should be known before the study.
investigator must have clearly defined null hypothesis in hand.
Margin of error
Acceptable significance level
It is a random sampling error, which is a likelihood of sample
it is in continuation with the previous line such as, (it is results variation from the population. For example, suppose there
denoted by a and acceptable level for significance means is 40% prevalence of anemia study sample and we set margin of
when a really true null hypotheses is rejected, in other words error as 5%; it means that range of anemia in population would
a = P [Type -I error]). Conventionally, 5% (α/P = 0.05) or be between 40 ± 5, i.e., 35% and 45% prevalence of anemia.
1% (α/P = 0.001) level of significance is considered by
the biomedical researchers; which means that research Standard deviation (SD) in the population
accepts that there could be 5% or 1% probability that results A researcher must anticipate the population variance of given
observed are due the chance not by our intervention. The outcome variable that may be calculated by mean that of SD.
resembling confidence levels (confidence interval [CI]) for For homogenous population, smaller sample size will be
the appropriate level of significance are: (a) CI-95% for the needed as variance or SD will be less in this population. For
5% (α/P = 0.05) level of significance and (b) CI-99% for the example, for studying the effect of exercise regimen on blood
1% (α/P = 0.001) level of significance. glucose, we include a population with blood glucose ranging
from 150 to 350 mg/dl. Now, it is simple to understand that we
Study power might require more number of samples to find out differences
It means possibility to rule out a significant difference when it among interventions because SD in this group will be more.
really exists. In other words, it is a probability of generalization Although if we consider a sample from population with blood
of study findings to population at large. An increase in sugar reading in between 150 and 250 mg/dl, then researcher
statistical power will decrease the possibility of Type II error may receive a more similar group representing homogeneity,
(β) occurrence; means reduces risk of false-negative results. therefore, decreases SD and number of samples for study.
Therefore, it is denoted as 1-β. In most of clinical trial the
power of 0.8 (80%) or greater is considered more appropriate One tail and two tail inferential statistical test
to find out a statistically significant difference. Power of 80% The choice of one-tail or two-tail test depends on the objective
means there are 20% chances that we may fail to identify a of the study. Research has a hypothesis that a new drug is more
significant difference even though it really exists. effective in reducing the blood pressure; then the one-tail test
could be sufficient to test the hypothesis, but if not sure that
Expected effect size the new drug may be more or less effective in lowering BP
It refers to the magnitude of the relationship between two as compared to existing drug then it is always better to use
variables as it occurs in population. For example, average two-tail test. Inputs for the one-tail and tow tail tests are same
hemoglobin rise with one type of diet is 5 g/dl and another is except the critical ratio (Z value); which is different in one-tail
2 g/dl; then, the absolute effect size would be 5–2 = 3 g/dl. Thus, test (Z1-α) and two-tail test (Z1-α/2) as depicted in Table 1.
effect size is considered as the calculated difference between
the measured effects of interventional and control group. Pilot Design effect (DEFF)
studies and previously reported data can be used to estimate The sample size calculation formulas provided in this article
the effect size. Cohen guide for effect size which is preferred helps to estimate an adequate sample size when simple random
by many social scientist states that it is considered to be small
effect if effect size is <0.1, effect size between 0.3 and 0.5 is Table 1: Z values
assumed to be medium effect and more than 0.5 is called as large Level of confidence (%) Two‑tailed (z1−α/2)
difference effect. Hence, effect size of 0.5 is commonly used as 0.05 (95) 1.96
it comes in reflects moderate to large difference. Nevertheless, 0.01 (99) 2.58
it is important to mention that effect size and sample size are
0.001 (99.9) 3.29
inversely proportional to each other; when effect size is large
Power of the test Z value (Z1−β)
research needs smaller sample size and vice versa.
0.80 0.84
Underlying event rate in the population 0.90 1.28
0.95 1.65
It is very essential to consider a prevalence rate or bottom
0.99 2.33
line event rate of the condition under study population

2020 | Vol 10 | Issue 01 National Journal of Physiology, Pharmacy and Pharmacology  2

Sharma et al. Calculation of sample size in nursing research

sampling technique is assumed to be used in the study. However, Sample size (n) = 406
whenever simple random sampling is not possible to use, then Z1-α/2 = 1.96
calculated sample may not be adequate and to overcome this P = 40% = 0.4
problem, calculated sample size has to be adjusted in terms of q = 1−0.4 = 0.6
DEFF. This is equal to the ratio of expected variance in cluster d = 5% = 0.05
random sampling with expected variance in simple random
sampling. The DEFF is generally ≥1. Therefore, in cluster Hence, for conducting a new cross-sectional study to identify
design, we assume DEFF = 2. the prevalence or proportion of DM among patients, minimum
406 subjects will be required.
DEFF = 1+δ(n−1) (Hers, δ = interclass correlation;
n = common size of the cluster) Example II: Sample size, when mean is of the study or data
are on interval/ratio scale
EXAMPLES FOR SAMPLE SIZE ESTIMATION (Z1-α /2 ) 2 * (σ ) 2
Sample size (n ) =
FOR OBSERVATIONAL AND EXPERIMENTAL ( d )2
STUDIES[11-23] n = Desired number of samples
z1-α/2 = 
Standardized value for the corresponding level of
Estimation of Sample Size for Cross-sectional or confidence.
Descriptive Research Studies (At 95% CI, it is 1.96 and at 99% CI or 1% type I error it is
These studies or surveys are generally conducted to find out, 2.58)
observe, describe, and document aspects of a situation as it d = Margin of error or rate of precision
naturally occurs. It is not used to identify the causation of σ = SD which is based on previous study or pilot study
something, such as a reason of any epidemic. Researchers do
not manipulate variables. A researcher might collect cross- Suppose a researcher wants to know the average hemoglobin
sectional data on past alcohol habits and current diagnoses of level among adults in the city at 95% CI and the margin of
liver disease, for example. error is 2 g/dl. From a previous study, the SD of hemoglobin
level among adults was found to be 4.5 g/dl. How many study
Example I: Sample size in case data is on nominal/ordinal subjects will be required to conduct a new study?
scale and proportion is one of the parameters
(z1− α / 2 ) 2* (p) (q ) On applying:
Sample size ( n ) = (Z1-α /2 ) 2 * (σ ) 2
(d) 2 Sample size (n ) =
n = Desired sample size ( d )2
Z1−α/2 = 
Critical value and a standard value for the (1.96)2 * (4.5)2
corresponding level of confidence. (n ) =
(At 95% CI or 5% level of significance (type-I error) it is (2)2
1.96 and at 99% CI it is 2.58) (n) = 19.44
P = Expected prevalence or based on previous research (n) = 19+2 (considering 10% dropout of study participants)
q = 1-p Sample size (n) = 21
d = Margin of error or precision z1-α/2 = 1.96
σ = 4.6 g/dl
A researcher wants to carry out a descriptive study to understand d = 2 g/dl
the prevalence or proportion of diabetes mellitus among adults
in a city. A previous study stated that diabetes in the adult Hence, for conducting a new cross-sectional study to estimate
population was 40%. At 95% CI and 5% margin of error, the average hemoglobin level among adults, minimum 21
calculate the sample required to conduct other new research? subjects will be required.

On applying: Example III: Sample size for finite population

(z1− α / 2 ) 2* (p) (q ) Sample required (n) = N/1+N*d2

Sample size ( n ) =
(d) 2
N = Total population
(1.96)2 * (0.4)(0.6)(0.05)2 d = Margin of error or precision
(n ) =
(0.05)2
(n) = 368.79 Suppose a researcher wants to conduct a survey to assess the
(n) = 369+37 (considering 10% dropout of study participants) prevalence of regular foot care among adults with diabetes

3 National Journal of Physiology, Pharmacy and Pharmacology 2020 | Vol 10 | Issue 01

Sharma et al. Calculation of sample size in nursing research

in a town. Now, if the population of adults suffering from On applying:

diabetes is 2500 and researcher wants to work at 95% CI,
where margin of error or precision rate decided by researcher (r + 1) (
p (1 − p ) Z1-β +Z1− α / 2 )
2

is 5%. Find out the sample size for the upcoming survey? Sample size ( n ) = *
r (p1 − p2 )2
On applying: Above formula (1 + 1) 0.35 (1 - 0.35) (0.84 + 1.96)
2

(n ) = *
(n) = 2500/1+2500*(0.05)2
1 (0.40 − 0.30)2
(n) = 2*178.36
(n) = 344.82 (n) = 356.72
(n) = 345+34 (considering 10% dropout of study participants) (n) = 357+36 (considering 10% dropout of study participants)
Sample size (n) = 379 Sample size (n) = 393
N = 2500 p1 = 40%=0.4
d = 5% = 0.05 p2 = 30%=0.3
r=1
Hence, for conducting a new cross-sectional study to assess Z1-β = 0.84
the prevalence of regular foot care among adults with z1-α/2 = 1.96
diabetes, minimum (379 subjects) will be required. p = 0.4+0.3/2 = 0.35

Therefore, a researcher is supposed to take a minimum of 393

Sample Estimation for Case–control Studies subjects in case as well as in the control group.
It is a study that determines the cause and effect to see
whether exposure is correlated with an outcome or not. By Example II: Sample size in case data is on interval/ratio
way of explanation, it determines wherever an exposure (quantitative) scale and mean as a parameter of the study
is correlated with an outcome (i.e., disease or condition ( )
2
(r + 1) σ 1−β + Z1− α / 2
2 Z

of interest). It is a type of observational study in which Sample required (n ) = *

r d2
commonly assumed causation is studied among two groups
differing in outcome. For example, a case–control research n = Number of samples which we need to find out
to find out the relationship between alcohol and liver disease. r = Control to cases ratio
p = Proportion of population = P1+P2/2
Example I: Sample size, when proportion is parameter of the
Z
1-β
= It is the desired power (0.84 for 80% power and 1.28
study or data are on nominal/ordinal scale: for 90% power)
z1-α/2 = Critical value and a standard value for the corresponding
(r + 1) (
p (1 − p ) Z1-β +Z1− α / 2 ) level of confidence.
2

n= * (At 95% CI it is 1.96 and at 99% CI or 1% type I error it is

r (p1 − p2 )2 2.58)
n = Desired number of samples σ = SD which is based on a previous study or pilot study
r = Control to cases ratio (1 if same numbers of subject in d = Effect size (difference in the means from previous studies
both groups) or pilot study)
p = Proportion of population = (P1+P2)/2
Z1-β = It is the desired power (0.84 for 80% power and 1.28 Suppose a researcher wants to conduct a study to identify
for 90% power) the association between the amount or quantity of alcohol
z1-α/2 = Critical value and a standard value for the corresponding consumption and liver cirrhosis. From the previous study,
level of confidence. he finds that the mean difference in alcohol consumption
(At 95% CI or 5% type I error it is 1.96 and at 99% CI or 1% between the case and control groups was 10 ml/day and SD
type I error it is 2.58) was 18 ml/day. He decides to conduct the study at 95% CI
P1 = Proportion in cases and fix power of the study at 80%. Find the sample size to
P2 = Proportion in controls have a equal number in case and control group for study.

If a researcher wants to conduct a case–control design to On applying:

identify the link between deep vein thrombosis and pulmonary
( )
2
embolism. He decides to work at 95% CI and 80% power of (r + 1) σ Z1-β +Z1− α / 2
2

n= *
the study. He assumes expected proportion in case is 40% and r d2
control group is 30% and decides to have a same number of
(1 + 1) (18) (0.84 + 1.96)
2 2
cases in both groups. Find out the optimum sample size for (n ) = *
each group in study. 1 10

2020 | Vol 10 | Issue 01 National Journal of Physiology, Pharmacy and Pharmacology  4

Sharma et al. Calculation of sample size in nursing research

(n) = 2*25.40 [1.96 √ {(1 + 1 / 1) 0.25 (1 - 0.25)} + 0.84

(n) = 51+5 (considering 10% dropout of study participants)
√ {0.2 * (1 - 0.2 / 1) * 0.3 (1 - 0.3)}]2 (0.2 - 0.3)
2
Sample size (n) = 56
(n) =
r=1 (0.2 - 0.3)2
Z1-β = 0.84
z1-α/2 = 1.96
(n ) =
[1.96 * 0.612 + 0.84 * 0.1833]
2

σ = 18 0.01
d = 10
(n) = [1.199+.1539]2/0.01
Therefore, a researcher will require minimum (56 subjects) in
the case as well as in the control group. (n) = 183.17
(n) = 183+18 = 201 (considering 10% dropout of study
participants)
Same Size Estimation for Cohort Studies Sample size (n) = 201
It is defined as a longitudinal research study that includes a r=1
category of people who share the same characteristic, typically Z1-β = 0.84
those who experienced a common event in a selected period, z1-α/2 = 1.96
such as disease or education. There is no control group, and p0 = 0.2
no intervention or treatment is given to patient that is why p1 = 0.3
it is different from randomized control trials. For example, m=1
researcher asks study subject to record their eating practices 0.3 + 1 × 0.2
P= = 0.25
over the period of time and then correlates between eating 1+1
practices and their sleep pattern. Therefore, a researcher will require a minimum (201 subjects)
for the study.
Example I: Sample size estimation for independent cohort
studies Sample size estimation for comparative studies
[ Z1− α / 2 √ {(1 + 1 / m ) p * (1 − p )} + Z1−β It is the study design in which comparison is done between two or
more groups on the basis of selected attributes such as knowledge,
n=
{
√ p0 * (1 - p0 / m ) p1 (1 - p1 ) ]2 } perception, and attitude. A multidisciplinary approach is best
(p0− p1 ) used for this type of researches. In case of comparative studies,
2

quantitative data of secondary analysis are outspread.

n = Total number of desired study subjects (case) to identify
true relative risk with two-sided Type-I error Example I: Sample size in case data is on nominal/ordinal
m = Number of subjects (control) per experimental subject scale and proportion is parameter of the study
Z
= It is the desired power (0.84 for 80% power and 1.28
1-β
for 90% power) p1 (1 - p1 ) + p 2 (1 - p 2 )
Sample size ( n ) = *C
z1-α/2 = Critical value and a standard value for the corresponding (p1 - p2 )2
level of confidence. n = Sample size for one group that we need to find out
(At 95% CI it is 1.96 and at 99% CI or 1% type I error it is 2.58) p1 and p2 = Proportion of two groups
p0 = Possibility of event in controls C = Standard value for the corresponding level of α and β
p1 = Possibility of event in experimental selected for the study. It is as follows:
p = p1+m p0/m+1
z1‑α/2 0.05 0.01
A researcher wants to identify the impact of smoking on lung Z1‑β 0.8 7.85 11.68
cancer. A previous study stated that proportion of lung cancer 0.9 10.51 14.88
in the case group is 30% and in the control group is 20%.
Calculate the sample size, if a researcher wishes to conduct A researcher wants to compare the knowledge among
the study at 95% CI and 80% power of the study with the Group A and Group B. From the previous study, proportion
equal number of case and control subjects. of the two groups is taken, i.e., 40% and 20%, respectively.
On applying: A researcher wants 95% CI and 80% power for the study.
Calculate the total subjects required for a new study.
[ Z1− α / 2 √ {(1 + 1 / m) p (1- p)} + Z 1− β On applying:

(n ) =
{
√ p0 * (1 - p0 / m ) p1 (1 - p1 ) ] } 2
p1 (1 − p1 ) + p 2 (1 − p 2 )
Sample size ( n ) = *C
(p0− p1 ) (p1 − p2 )2
2

5 National Journal of Physiology, Pharmacy and Pharmacology 2020 | Vol 10 | Issue 01

Sharma et al. Calculation of sample size in nursing research

0.4 (1 − 0.4) + 0.2 (1 − 0.2) σ = Common SD

(n ) = * 7.85 c = Constant value depends on the value of α andβselected for
(0.4 - 0.2) 2

the study. It is as follows:

(n) = 10*7.85
(n) = 78.5+7 (considering 10% dropout of study participants) z1−α/2 0.05 0.01
Sample size (n) = 86 Z1−β 0.8 7.85 11.68
p1 = 40%=0.4 0.9 10.51 14.88
p2 = 20%=0.2
C = 7.85 (at 95% CI and 80% power) A researcher wants to know the effect of a drug A and compare
the drug A with placebo. He thinks that if drug A decreases
Therefore, a researcher will require minimum (86 subjects) the blood sugar level by 15 mg/dl as compared to placebo,
for the study in each group. then it should be considered as clinically significant. Suppose
a previous study stated that the SD was 30 mg/dl. He decided
Example II: Sample size in case data is on interval/ratio scale to conduct the study at 95% CI with 80% power of study.
and mean is parameter of the study
On applying:
( )
2
(σ +σ 2 ) Z1-β +Z1-α /2
2 2 2

Sample size (n ) = Sample size (n) = 1+2C (SD/d)2

1

(d ) 2

(n) = 1+2*7.85 (30/15)2

d = difference in means of two group (effect size) (n) = 1+62.8
σ1 = SD of Group 1 (n) = 63.8+6 (considering 10% dropout)
σ2 = SD of Group 2 Sample size (n) = 70
Z
1-β
= It is the desired power d = 15 mg/dl
z1-α/2 = Critical value and a standard value for the corresponding SD = 30 mg/dl
level of confidence. C = 7.85 (at 95% CI and 80% power)
(At 95% CI it is 1.96 and at 99% CI, or 1% type I error it is 2.58)
Therefore, researcher will require minimum of 70 subjects
Assuming from the previous study that the mean (SD) of for the study in each group that means a total of 140 study
HbA1c level in Group A and Group B was 5.5 (2.2) and subjects are required.
7.5 (2.0), respectively. At 95% CI and 80% power of the
study, find out the sample size for the new study.
Sample Size Estimation for Experimental Studies

On applying: Experimental studies or randomized controlled trials are the

studies in which researcher artificially manipulates variables
( ) under the study. Randomization and control group are an important
2
(σ 12 + σ 2 2 )2 Z1− β + Z1−α / 2
Sample size (n ) = aspect in these types of studies. In this investigator provides
d2 intervention and study its effect and compare in experiential and
(2.22 + 22 )2 (0.84 + 1.96)
2 control group. There are following types of comparison:
(n ) = • Superiority trial: The aim of this type of trial is to view
22 that a novice drug/treatment/intervention is superior to a
(8.84)2 (2.8)2 control treatment.
(n ) = • Equivalence trial: The aim of this type of trial is to view
4
that novice treatment/intervention is equally effective to
(n) = 153.15+15 (considering 10% dropout) control treatment.
Sample size (n) = 168 • Non-inferiority trial: The aim of this type trial is to view
that novice treatment/intervention is effective but must
Therefore, a researcher will require a minimum (168 subjects) not superior than the control treatment.
for the study in each group that means a total of 336 study
subjects are required. Example I: Sample size to rule out the difference (effect size)
among two groups (on the basis of difference in proportion or
Example III: Sample size for comparison between two groups for dichotomous nominal/ordinal variables)
and for continuous variables
n = Sample size for each group
Sample size (n) = 1+2C (SD/d) 2 d = Difference in means of two treatment effect
Zx = Standard value for a one or two-tailed
n = Sample size for one group that we need to find σ2 = SD of Group 2
d = Detected difference in means of two group (effect size) 𝛿0= Acceptable margin of error

2020 | Vol 10 | Issue 01 National Journal of Physiology, Pharmacy and Pharmacology  6

Sharma et al. Calculation of sample size in nursing research

S2 = Pooled SD (both comparison groups) Example II: Sample size to rule out the difference (effect size)
p = Response rate of standard intervention among two groups (on the basis of difference in the mean or
p0 = Response rate of new intervention for continuous variables).

A researcher wants to assess the difference in the effectiveness A researcher wants to see the variation in the effectiveness of
of treatment A (new intervention) and treatment B (standard drug-X (new drug) and drug-Y (standard drug) prescribed for
intervention) for the treatment of stroke in two-month protocol. diabetes treatment. Change in blood glucose level (mg/dl) is
the primary outcome, compared to values at baseline study
Researcher assumes all parameters as: point at 95% CI and 80% power of the study. The researcher
assumes all parameters as:
p = 0.30; p0=0.45; a=0.05 (95% CI); β=0.20 (80% power);
Z1-α = 1.645, Z1-β=0.84, Z1-α/2=1.96
d=0.16; 𝛿=0.21, 𝛿0=0.10
• Mean change of blood glucose level in the new drug
a. Non-inferiority trial:
treatment group is 20 mg/dl
 Z1-β + Z1-α 2 p  • Mean change of blood glucose level in the standard
Sample size ( n ) = 2 *   * (1 − p )
 δ0  treatment group is 16 mg/dl
• S (pooled SD of both comparison groups) = 10 mg/dl
 0.84 + 1.6452 0.3  • 𝛿 (actual margin between two intervention) = 4 and 𝛿0
n = 2*  * (1 - 0.3)
 0.10  (clinically allowable difference) = 2.
n = 260.40+26 (considering 10% dropout)
Sample size (n) = 286 a. Non-inferiority trial:
p = 0.30  Z1−β + Z1− α 
2

Z1-α =SD=1.645 Sample size ( n ) = 2 *   * SD

2

 δ0 
Z1-β = 0.84
𝛿0 = 0.05  0.84 + 1.6452 
(n ) = 2 *   * 10
2

 2 
b. Equivalence trail
(n) = 308.7+30 (considering 10% dropout)
 Z1-β + Z1− α / 2 2  Sample size (n) = 339
Samplesize ( n ) = 2 *   * p * (1 − p )
 δ0 
b. Equivalence trial
 0.845 + 1.962 
n = 2*  * 0.3 * (1 − 0.3)  Z1−β + Z1− α / 2 
2

 0.1  Sample size ( n ) = 2 *   * SD

2

 δ0 
n = 330.45+33 (considering 10% dropout)
Sample size (n) = 363  0.845 + 1.962 
n = 2*  *10
2
p = 0.30  2 
Z1-α/2 = 1.96
n = 280.5+28 (considering 10% dropout)
Z1-β = 0.84
Sample size (n) = 308
𝛿0 = 0.1
c. Clinical superiority trial:
c. Clinical superiority trial
 Z1−β + Z1− α 
2

 Z1-β + Z1− α 
2
Sample size ( n ) = 2 *   * SD
2

Sample size ( n ) = 2 *   * p * (1 − p )  δ − δ 0 
 δ − δ 0 
 0.84 + 1.6452 
 0.84 + 1.6452  (n ) = 2 *   *10
2

(n ) = 2 *   * 0.3 * (1 − 0.3)  4−2 

 0.21 − 0.1 
(n) = 248.5+24.8 (considering 10% dropout)
(n) = 2*510.34*0.3*0.7 Sample size (n) = 273.
(n) = 214+21 (considering 10% dropout)
Sample size (n) = 235
p = 0.30 Electronic Resources for Sample Size Calculation
Z1-α = 1.645 Today, there are many sample size calculation software
Z1-β = 0.84 or internet URL links available. Some of the commonly
𝛿0 = 0.10 used sample size calculation electronic resources are given
𝛿 = 0.21 below:

7 National Journal of Physiology, Pharmacy and Pharmacology 2020 | Vol 10 | Issue 01

Sharma et al. Calculation of sample size in nursing research

S. No. Software URL Link

1. G Power* http://www.ats.ucla.edu/stat/gpower/pairedsample.htm
2. Raosoft Software http://www.raosoft.com/samplesize.html
3. Epi Info https://www.cdc.gov/epiinfo/support/tutorials.html
4. PASS Sample Size Software https://www.ncss.com/videos/pass/
5. IBM SPSS https://www.ibm.com/support/knowledgecenter/en/SSCVKV_10.0.0/Campaign/CampaignProcesses/
To_use_the_sample_size_calculator.html
6. Growth Analyzer https://growthanalyser.org/software/growth‑analyser‑rct/
7. Java Applets https://homepage.divms.uiowa.edu/~rlenth/Power/

CONCLUSION In: Applied Health Research Manual: Anthropology of Health

Care. Amsterdam: Het Spinhuis; 2001.
This article has provided a simple and comprehensive 13. Kadam P, Bhalerao S. Sample size calculation. Int J Ayurveda
discussion about sample size calculation for the observational Res 2010;1:55-7.
and experimental nursing studies, which will enable nurse 14. Noordzij M, Tripepi G, Dekker FW, Zoccali C, Tanck MW,
Jager KJ, et al. Sample size calculations: Basic principles and
researcher to identify the appropriate sample size for their
common pitfalls. Nephrol Dial Transplant 2010;25:1388-93.
studies to produce more reliable results to enhance the
15. Dell R, Holleran S, Ramakrishnan R. Sample size determination.
generalization of study findings. ILAR J 2002;43:207-13. Available from: https://www.ncbi.nlm.
nih.gov/pubmed/12391396. [Last accessed on 2019 Mar 05].
REFERENCES 16. Zhong B. How to calculate sample size in randomized
controlled trial? J Thorac Dis 2009;1:51-4.
1. Chan YH. Randomised controlled trials (RCTs) sample size: 17. SchallR.Bookreview:Testingstatisticalhypothesesofequivalence.
The magic number? Singapore Med J 2003;44:172-4. Clin Trials 2004;1:139-40. Available from: https://www.journals.
2. Charan J, Saxena D, Yadav P, Kantharia ND. Negative studies sagepub.com/doi/abs/10.1191/1740774504cn012xx.[Last
published in medical journals of India do not give sufficient accessed on 2019 Mar 05].
information regarding power/sample size calculation and 18. Hwang I, Morikawa T. Design issues in noninferiority/
confidence interval. J Postgrad Med 2011;57:176-7. equivalence trials. Drug Inf J 1999;33:1205-18. Available from:
3. Charan J, Yadav P, Kantharia ND. Reporting of sample size and https://www.pdfs.semanticscholar.org/ 9b06/ 7c2d18513b1051
power in negative clinical trials published in Indian medical 6743a16fd2f12711166714.p. [Last accessed on 2019 Mar 05].
journals. J Pharm Negat Results 2011;2:87-90. 19. Bujang MA, Adnan TH. Requirements for minimum sample
4. Jayakaran C, Saxena D, Yadav P. Negative studies published size for sensitivity and specificity analysis. J Clin Diagn Res
in Indian medical journals are underpowered. Indian Pediatr 2016;10:YE01-6.
2011;48:490-1. 20. Carley S, Dosman S, Jones SR, Harrison M. Simple nomograms
5. Naduvilath TJ, John RK, Dandona L. Sample size for to calculate sample size in diagnostic studies. Emerg Med J
ophthalmology studies. Indian J Ophthalmol 2000;48:245-50. 2005;22:180-1.
6. Patra P. Sample size in clinical research, the number we need. 21. Mishra P, Pandey MP, Singh U, Sharma V, Yadav SS, Kar R.
Int J Med Sci Public Health 2012;1:5-9. Sample size estimation for clinical research studies using mean
7. Shah H. How to calculate sample size in animal studies? Natl J and proportion. Int J Sci Res 2017;6:587-79.
Physiol Pharm Pharmacol 2011;1:35-9. 22. Sakpal TV. Sample size estimation in clinical trial. Perspect
8. Malterud K, Siersma VD, Guassora AD. Sample size in Clin Res 2010;1:67-9.
qualitative interview studies: Guided by information power. 23. Charan J, Biswas T. How to calculate sample size for different
Qual Health Res 2016;26:1753-60. study designs in medical research? Indian J Psychol Med
9. Kirby A, Gebski V, Keech AC. Determining the sample size in 2013;35:121-6.
a clinical trial. Med J Aust 2002;177:256-7.
10. Cohen J. Statistical Power Analysis for the Behavioral How to cite this article: Sharma SK, Mudgal SK, Thakur K,
Sciences. 2nd ed. London: Academic Press; 1977. Gaur R. How to calculate sample size for observational and
11. Shankar P. Essential medicines and health products information experimental nursing research studies? Natl J Physiol Pharm
portal. J Pharmacol Pharmacother 2014;5:75. Available from: Pharmacol 2020;10(01):1-8.
http://www.apps.who.int/medicinedocs/pdf/s6169e/s6169e.
pdf. [Last accessed on 2019 Mar 05].
Source of Support: Nil, Conflicts of Interest: None declared.
12. Hardon A, Boomongkon P, Tan E. Analysis of qualitative data.

2020 | Vol 10 | Issue 01 National Journal of Physiology, Pharmacy and Pharmacology  8

View publication stats