http://makemy2day.blogspot.com/

visit http://makemy2day.blogspot.com/
for sample protocols of my students

How to validate a questionnaire?

How to validate a questionnaire?

 

http://pediatricthesis.blogspot.com/2018/ 

[12/05, 10:25 PM] Santosh Kondekar: How many questions are there
And how many components are there for the questionaire
What kind of answer patterns these questions have
R u planning a scoring system too
[12/05, 10:50 PM] Santosh Kondekar: In simple words
If its a serious questionaire study u need to do component validation, so that further study doesnt go waste. Each component and pattern of questions and or score needs to b derived without ambiguous results and avoiding bias and errors of random filling of questionnaire.
If there is yes no question it becomes single question, if it has many options it becomes many questions... Like a mcq.
Unless we confirm the questionaire is sound for a random accessor by doing some pilot filling and analysing result and modify as per needs felt in that pilot study, there will b plenty of revisions in it.
So only when the questionare iw completely sound as per u, u can subject it for validation. Simplest methodfor validation is called Heuristic method or face validation by which u get ur question Aire approved and corrected and okayed by 5-10 different people or specialist working in same field. Each person validating is supposed to check components construct and biases
[12/05, 10:50 PM] Santosh Kondekar: After its validated .. the questionaire is considered primary validated.
[12/05, 10:50 PM] Santosh Kondekar: https://www.statsdirect.com/help/basics/validity.htm also read this
[12/05, 10:52 PM] Santosh Kondekar: Depending on prevalence decided sample size, the questionare is then applied to the given population and quantitative results of study are analysed as per assumptions of aims n objectives for different components. When assumptions turn true by stats tests, its considered externally validated.

profile of case studies: statistical analysis possibilities

for most studies that assess profile The results will expressed as mean and standard deviation values or median and range for interval variables and as proportions for categorical and ordinal variables. Correlation of interval variables will be analyzed using Pearson’s correlation test. If normal distribution could be assumed, comparison of interval variables between two or more independent samples will be analyzed using the independent t-test or one-way ANOVA, respectively. Interval variables without normal distribution, ordinal and categorical variables will be analyzed using the Wilcoxon-Mann-Whitney or Kruskall-Wallis tests when appropriate. Proportions will be analyzed using Pearson's chi-square test; Fischer’s exact test will be used as an alternative in contingency tables of 2x2 variables. A p-value of <0.05 will be considered significant; when using sequential Mann-Whitney tests, p-value significance was calculated using the Bonferroni correction. and for prevalence studies: 1. prevalence is expressed as percentage studies with numertaor and denominator 2. incidence may be expressed with newly diagnosed cases in numerator 3. Subgroups may be studied with modified Pearsons T test 4. If the sample size more than 500, Spearman test may be applicable between subgroup correlations. Kondekar

Dr kondekar is reading these days about

View my collection, "My Bibliography" from NCBI

thesis topics: clinical profiles: descriptive studies

clinical profile studies are descriptive studies. groups numbers percentages. no heavy statistics; usually for a new pattern search or when you have nothing in mind and your guide or EC creates obstacles for RCT or novel ideas.
clinical profile of cases  with ..........................................
write any commonly seen disease, morbidity, sign, symptom, report etc

example
clinical profile of patients presenting with non watery loose motions
clinical profile of dengue like patients with tests negative
clinical profile  of patients with SAM in hospitalized patients
clinical profile of patients with SGOT>PT ratio > 1

as its an observational study, its u who has found something new.. and want to know whats happening.

Dr kondekar

MD/DNB/UG MBBS pediatric icmr project/ thesis topics: how to decide, what to decide?

lets be more practical .
for a 3 year MD course say in pediatrics, you need to have 2 years for study and at least 1 year around the topic ( though its supposed to be throughout course)
first 6 months: basic reading and topic submission to EC
CLICK HERE TO REGISTER FOR  GCP TRAINING CERTIFICATE FOR FREE

next 1 year project completion, then analysis/results/submission etc
================================
process: Please read rules of ECARP  or ethics comittee before u prepare draft.
1. decide a topic
2.check feasibility
3.check similar studies
4.understand what and why
5.prepare a protocol as per needs of local ethics committee or ECARP
6. protocol should have a title page, index page and drafts for each of the following headings: on separate page. The headings are title, introduction, rationale, aims and objectives, inclusion criteria, exclusion criteria, methodology study type and  flow chart,  Study process, statistical plan for analysable variables as per the type of study, past research or review of literature references, limitations of the study and a one page protocol summary, case record form, consent /assent forms (ICF) ,Patient information sheet (PIS), in different languages, Bibliography, appendix, special permissions if any and master chart format if prepared
=================================

any delay in this will affect your exam reading/practice schedules.
difficult topics will require interventions follow up , more no of cases and may be finances so wont be easy for dissertations, though RCTs are best and easy to do, you need to select topics carefully, it should not only remain as a drug trial.
metaanalysis is not taken well by examiners as it doesnt involve direct patient contact.
Short term, easy topics, easy methods, easy but intellectual analysis: 
1. those who dont need an intervention like blood investigation or scan
2. doesnt involve lots of finances/sponsors
3. has no or short follow up
4. some one has previously worked on it
5. cases should be easy to pick up from every day OPD

best studies for such easy ones are done by
1. finding a study from last few journals or old thesis topics
2. zero down your area or subject of interest
3 look for some tool/scale/policy/guidelines/score, that may  help in validating tool , studying the application of tool and making the tool and study statistically stronger.
Teachers prefer their area of interest for study topics so think  than search regarding such topics too.
4 look for some day to day cases where you dont find much uniformity in management, search for studies in this regard.


statistics:
for descriptive study:
This is an observational study .No power calculations are involved. The study period will be for 6 months. For calculating percentage of use topic under study, total number of surgeries done during the same period will be taken as denominator. Data will be assessed for normality and presented as mean (sd), median, range, and Frequency. (percentage). Unpaired t test will be used for numerical data and Chi square test or Fischer Exact test ( for binary data) for categorical data.
Dr kondekar
also visit http://medicalthesis.blogspot.com