Slide 1

0 comments

file time: 2008-02-16

filetype:pptx

Click Here To Download...

>  

BNURS

Statistics for Health Research 
Basics 

Graham J. Patrick Ph.D., ARNP

 

What is data & why do we collect it?

 

Common Sense Inquiry involves: 

Concepts are loosely joined and subjective Testing is subjective and selective Control of extraneous variables is usually not important Relationships are ignored if incongruent with beliefs Explanations are usually subjective and not necessarily based on facts  

Scientific vs Common Sense 
Modes of Inquiry

 

Scientific Inquiry involves: 

Concepts are systematically and objectively

                    developed and tested

Testing is systematic, controlled & empirical Control of extraneous variables is important Relationships are examined for exceptions Explanations are based on observed phenomena

                           and known facts

 

Important Considerations 

What is your view of Reality and Truth?  
 
What is the question you seek to answer?  
 
What kinds of data do you plan to collect?  

Types of Research 

Qualitative (subjective lived experiences of participants and may include those of the researcher) Quantitative (facts and numbers and the researcher is objective and impersonal)  

So How Do We Decide Which Approach to Use ? 

Qualitative 

Quantitative

 

Significance? 

Time 1                           Time 2

110/70                            100/70

135/100                          120/100

150/95                            110/70

120/75                            119/75

100/70                            110/72

170/110                          150/112

220/110                          150/100

150/100                          120/95

 

Qualitative 

How do participants view their BP? What do they say about their BP? What is the shared versus individual meaning of BP as a concept?  

Quantitative 

What is the average BP of subjects in this study before and after intervention? What is the range of blood pressures before and after intervention in this study? What variables affect blood pressures in this study?  

Important Terms 

Correlation-the degree of association between two items

                     of data

Data-information that is collected and pertains to some

          question of interest

Descriptive statistics-methods of classifying or

                                       summarizing data

Design-a selected process for uncovering relationships in

              data

Frequency distribution-a representation of data that

                                           portrays the occurrence of values 

 

More Terms 

Hypothesis-predicting an expected relationship between

                      variables

Inferential statistics-procedures for making

                                      generalizations about a population

Mean-the mathematical average of scores; a measure of

             central tendency

Median-the midpoint or middle score in a set of data

Modality-the number of peaks in a frequency distribution

 

More Terms 

Modal percentage-the percent of cases/scores in the mode

Mode-the most frequent score in a data set; a measure of

           central tendency

Multiple regression-the relationship between independent

                                      and dependent variables

Null Hypothesis-usually means no relationship between

                               group means 

 

More Terms 

Parameter-a measurement that describes a

                   population value

Population-all members of a specified group

Probability-the repeatability of an event in trials

                     under similar circumstances

Sample-a collection of data from a population 

 

More Terms 

Standard Deviation-average deviation of scores from the

                                    mean of a data set.

Statistic-a descriptive measure from a sample

Statistical significance-extent to which results can be

                                      attributed to chance

Statistics-the theory and methods applied to

                 understanding data

 

More Terms 

Type 1 error-inappropriate rejection of the null

                       hypothesis

Type 2 error-inappropriate acceptance of the null

                       hypothesis

Variable-a characteristic of a group that has different         

                values

Variance-the average squared distance from the mean 

 

Populations and Samples 

Populations are studied in terms of parameters and the average is 渭 or Mu, standard deviation is 0 and 0 虏 for variance Samples are a sub-set of a population and are studied in terms of statistics and the average X bar or X , S for standard deviation and S虏 for variance  

Statistics 

Are classified as: Descriptive -describe relationships in the data  
Inferential- analyzes relationships within sample

                        data in order to generalize from a

                      sample to the population from which

                      it may have come

 

Scales 

Nominal (gender)  
Ordinal (educational levels  
Interval (temperature)  
Ratio (Height or weight)  

Scales 

Nominal is simplest level and nominal numbers are used to differentiate objects.  For instance M= male, F= female; can also be 1=M or 2=F. This is used only to separate into categories.  We can tally numbers or do percentages, but not averages, or other statistical processes  
 
Ordinal is next more complex and involves ordering or ranking and one can determine whether one object has more or less of measured quantity for instance age, education, income, etc.  One cannot do average (rank cannot be averaged)  

Scales 

Interval indicates distance, difference and direction as equal level measurements are used.  For example temperature 32  0vs 33  0 but the intervals cannot be divided and 0 on scale is arbitrary and doesn00 indicate an absence of the value (0  0 is still a temperature and varies with the scale used e.g Fahrenheit vs Celsius)  
 
Ratio is highest level, tells difference, direction, distance and has a zero point.  Thus one can divide 2 ration level numbers and turn them into a ratio.  Most  physical measurements such as height, weight, BP, speed, etc. are ratio level measures  

So What? 

The main point of all of the preceding is that one has to think through what kinds of measures one must make in order to be able to actually do the types of statistical measures one needs in order to compare data with norms, previously collected data or?  

The following is an example of an outcome study conceived, developed, and conducted by nurse researchers and funded by the National Institutes of Health Division of Ageing

 

Nursing Research in Action:

Relaxation Training for Alzheimer00

Caregivers 

Dr. Sharon Lewis Ph.D., FAAN

Dr. Dody Clough R.N., Ph.D

Dr. Graham J. Patrick Ph.D., A.R.N.P 

NIH grant # 85-6000-642

 

Why the interest in Caregiver stress? 

Many studies indicate stress & burden are

huge problems for family caregivers, Vitaliano

et.al (1989), Schultz et. al, (1990),  & Kuhlman et

al. (1991) 

Studies also indicate the stress of caregiving is

associated with health risks, Kiekolt-Glaser

(1987), & Pruchno et. al., (1990)

 

Who are the researchers? 

Dr, Lewis is an internationally renowned

Psycho-neuroimmunologist 

Dr. Clough00 specialty is Qualitative Research 

My specialty is bioinstrumentation & stress

research

 

Relaxation Training for Alzheimer00 Caregivers

is a 5 year grant designed to evaluate the

effects of teaching relaxation skills to

caregivers of patients with Alzheimer00 disease 

Aims were to increase quality of life and

Immuno-competence through stress reduction 

2 groups were used:

Group 1 had relaxation training via coaching and use of a relaxation tape;

Group 2 waits 2 months and had repeated stress and immune measures followed by coaching and use of a relaxation tape

 

Intervention 

A brief professional pre-programmed

relaxation tape was given to each participant 

They were instructed to listen to the tape at

least twice a day preferably during the day and

at bed-time 

They were also allowed to pick from 2 types of

tapes, one with music and one voice only

 

Measures

Pencil and papers tests: SOC, PRQ85, CRI,

PSI, SF-36, CES-D, Symptom Questionnaire;

qualitative measures of subjects00experiences;

Bioinstrumentation measures of temperature, HR,

SCL, BP, EMG, and breathing rate are obtained

during a pre-programmed stress test that lasts 14

minutes, followed by a relaxation period of 30

minutes.

Blood immune measures of  natural killer  cells are

also obtained

 

                 Caregiver Qualitative Questions 

1.  What currently causes the most stress in your life?

     Are these similar stressors to what you00e experienced

     in the past? 

2.  When you think about being stressed where do you

     physically feel your stress? 

3.  Think about your physical stress.  How would you

     describe it? 

4.  Think about your physical stress again.  When you

     experience this stress, how does it make you feel?

 

Bioinstrumentation Stress Test 

Subjects were connected to bio-sensors and to some instruments connected to a PC.

The process was explained and demonstrated.

They were then exposed to a pre-programmed 18 minute stress test. They were encouraged to perform as quickly and as well as they are capable. 

The computer program guided them through the stressors via audio and visual prompts.

Following the stress period, subjects were told to relax as best they can for 30 minutes.

 

Bioinstrumentation Stress Test 

A catheter was inserted in the subjects00arm at the beginning of the stress test.  Blood samples were periodically drawn during the different aspects of the stress test and also during the relaxation period that follows. 

 

Bioinstrumentation Stress Test 

The stress test was approximately 18 minutes in length.  There are 4 stressors: Mental Math, Stroop, Cognitive Stressor, and Verbalized Stressor.  Prior to the 1st stressor and after each stressor there was a 2-minute rest period. 

The following slides are examples of the individual stressors in the test. 

Red Orange Yellow Green Blue Purple  

Stroop Word Color Test.  00ry to name  the color in each slide without reading the word.  You will only have a brief time so work as fast as you can00 00ORK FASTER!!!00/b>

 

Mental Arithmetic Stressor 

Start with 2000 subtract 17 and work as fast as you can. You have 2 minutes starting now WORK FASTER !!!!!  

Cognitive Stressors 

00or the next 2 minutes we would like you to

think about a recent stressful event.  Try to

capture it as real as possible. Try to capture

what happened, who was there, what you

thought, how you felt, and what it meant to

you00

 

Verbalized Stressor 

00ow for the next 2 minutes we would like you

to talk about the previous event in terms of

what happened, how you felt, and what it

Meant00

 

Relaxation Period 

00ow for the next 30 minutes we would

like you to relax as much as you can00  

***For the pretest subjects are instructed

to relax as they normally do and for the

post-test they are instructed to practice

the relaxation skills they have learned in

the program.

 

What we found 

The following 3 slides review the kind of data

we typically saw. 

Most subjects show a typical patterns of

increased stress response the first time they go

through the program.

 

Results 

The immune and bioinstrumentation

data indicated a clear pattern of

improvements following training. 

 

Results 

Caregiver objective burden seemed similar for Spouse and Children Caregiver subjective burden was less for spouses General health and mental health were both worse for spouses than children Anxiety, anxiety symptoms, and depression were all higher for children Overall quality of life was lower for children on most measures  

Results 

Natural killer cell number was higher for spouses than children Overall T-cell function was lower for spouses  
Overall T-cell counts showed  improvements over training and beyond  

Data Analysis 

We used P= < or = 0.05 as

Our acceptable probability of detecting a

true difference versus accepting a false

difference

There are also some inherent difficulties

in setting P values for some biofeedback

measures.  All measures aren00 equal in

terms of relative values of increases or

Decreases, or in how much of the value they had

to begin with.  In order to offset this you

need lots of subjects. 

 

Results 

For last relax period best data indicated the mean difference in subjects00EMG was significantly different (P= 0.034). 
 

Thoughts?

 

FIN

   download Slide 1