Literature DB >> 30112327

Cautious use of absolute and relative precision while calculating sample size.

Preeti Usha1, Ajeet Singh Bhadoria1, Surekha Kishore1.   

Abstract

Entities:  

Year:  2018        PMID: 30112327      PMCID: PMC6069639          DOI: 10.4103/jfmpc.jfmpc_39_18

Source DB:  PubMed          Journal:  J Family Med Prim Care        ISSN: 2249-4863


× No keyword cloud information.
Dear Editor, Recently published article entitled “Locomotor problems among rural elderly population in a District of Aligarh, North India” has referred an extremely topical subject of locomotor problems among the most vulnerable geriatric population residing in a rural area.[1] Authors have rightly targeted sociodemographic risk factors to document their relationship with the presence of locomotor problems among the elderly population. However, we have following concerns about this study and intend to highlight here to help readers to be cautious before drawing conclusions. Related to prevalence, the sample size calculated has some major concerns. It was calculated considering lowest prevalence and absolute precision. With any absolute precision, maximum sample size is attained with 50% prevalence.[2] The concept of lower the prevalence and higher the sample size is correct if relative precision is taken to calculate the sample size. Thus, the sample size of 225 was inadequate and study was underpowered to document the desired prevalence Authors have mentioned that a pilot study was done to get baseline information about the health problems. The findings of the pilot study should have been utilized to calculate the sample size using appropriate statistics The utilization of systematic random sampling to select 225 elderlies from total sampling frame of 1018 is well justifiable. However, how and why the probability proportionate to size (PPS) method was used is not clarified. PPS is used to select clusters in cluster sampling method.[3] If it is used, then authors should have shared the details of the clusters and should have considered the design effect in calculating the sample size to take care of intraclass correlation among the clusters Important sociodemographic components such as socioeconomic status including income and type of family in which elderlies were living and family size were not included in the study. Enough evidences are available to acknowledge their role with respect to locomotor and other disabilities among geriatric population[4] Multivariate regression analysis could have been utilized to document the adjusted odds ratio for working status after adjusting for the nonmodifiable significant risk factors such as age and gender Authors have excluded severely ill or moribund individuals which will lead to underestimation of prevalence of locomotor disability as the definition utilized itself includes those with loss or lack of normal ability to execute distinctive activities associated with day-to-day routine.[5] These people might also have locomotor and other problems related to old age. Thus, the prevalence might be underestimated, study looks underpowered with inadequate sample size, and sociodemographic risk factor profile was incomplete. Therefore, readers need to be careful to reach to any conclusion.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.
  2 in total

1.  A simplified general method for cluster-sample surveys of health in developing countries.

Authors:  S Bennett; T Woods; W M Liyanage; D L Smith
Journal:  World Health Stat Q       Date:  1991

2.  Locomotor problems among rural elderly population in a District of Aligarh, North India.

Authors:  Mohd Maroof; Anees Ahmad; Najam Khalique; M Athar Ansari
Journal:  J Family Med Prim Care       Date:  2017 Jul-Sep
  2 in total

北京卡尤迪生物科技股份有限公司 © 2022-2023.