Document(s) 11 of 27

Flowchart: Steps in the development of a health systems research proposal

NB: Development of a research process is a cyclical process. The double-headed arrows indicate that the process is never linear.

Module 8: VARIABLES

1. Introduction
2. Formulating variables
3. Identifying indicators in qualitative studies
4. Causes and associations; confounders

I. INTRODUCTION

In Module 4 we analysed the problem we wanted to investigate. The problem itself and all the factors that might influence it were presented in a diagram, which then served as the basis for the formulation of research objectives. Now we have come to a stage where we must ask ourselves the question:

‘What information are we going to collect in our study to meet our objectives?’

• In most studies, we must first describe the problem itself more precisely.

  For example, in a study that is investigating why so many tuberculosis (TB) patients default from out-patient treatment, we first want to know how high the defaulter rate is: is it 10%, 30%, 50%? To obtain the defaulter rate we need a clear definition of what we mean by defaulting (how many times treatment was missed).

• We also want to know whether certain factors do indeed influence the problem, and to what extent. If we know the extent to which a certain factor influences the problem, we are much more likely to be able to convince ourselves (and relevant others) to take action.

  For example, if we find that becoming a dropout of TB treatment is strongly associated with the following factors, we have clues that will help us to solve the problem:

  —The patient’s lack of knowledge concerning the actual duration of treatment and the danger of relapse or death when the full course is not completed;

  — Living more than 8 km away from the clinic where the drugs have to be collected monthly; and

  —Being between 15 and 30 years of age.

To find these associations between problems and contributing factors, it is essential that we carefully define the problem itself, as well as each of the factors identified when analysing the problem in Module 4. We do this by formulating variables.

II. FORMULATING VARIABLES

What is a variable?

A simple example of a variable in the form of numbers is ‘a person’s age’. The variable ‘age’ can take on different values since a person can be 20 years old, 35 years old and so on. Other examples of variables are:

• weight (expressed in kilograms or in pounds);
• home - clinic distance (expressed in kilometres or in minutes walking distance);
• monthly income (expressed in dollars, rupees, or kwachas); and
• number of children (1, 2, etc.).

Some variables may also be expressed in categories. For example, the variable sex has two districts categories, groups, male and female. Other examples are:

Table 8.1: Examples of categorical variables

Further breakdown of numerical and categorical variables (optional)

Numerical variables can either be continuous or discrete.

1. Continuous. With this type of data, one can develop more and more accurate measurements depending on the instrument used, e.g.:
   • height in centimetres (2.5 cm or 2.546 cm or 2.543216 cm)
   • temperature in degrees Celsius (37.2⁰C or 37.19999⁰C etc.)
2. Discrete. These are variables in which numbers can only have full values, e.g.:
   • number of visits to a clinic (0, 1, 2, 3, 4, etc).
   • number of sexual partners (0, 1, 2, 3, 4, 5, etc.)

Categorical variables, on the other hand, can either be ordinal or nominal.

1. Ordinal variables. These are grouped variables that are ordered or ranked in increasing or decreasing order:

   For example:

   High income (above $300 per month); Middle income ($100-$300 per month); and Low income (less than $100 per month).

   Other examples are:
   Disability:	no disability, partial disability, serious or total disability
   Seriousness of a disease:	severe, moderate, mild
   Agreement with a statement:	fully agree, partially agree, fully disagree
   Fear of leprosy:	will not share food with a patient; will not enter the house of a patient; will not allow patient to live in the community.

   Note:

   Fear of leprosy is an attitude, and attitudes are often scaled (you make them into ordinal variables).

   It is obvious that the definition of what we would call high (income) or far (distance) will vary from country to country and from region to region. If a researcher has little idea about the distribution of a certain variable in a population (for example, if you don’t know whether 30%, 50%, or 95% are below the poverty line of $100 per month), it is advisable to categorise numerical data only after the pre-test, or even after data collection (see Module 13).

2. Nominal variables. The groups in these variables do not have an order or ranking in them.

   For example:
   Sex:	male, female
   Main food crops:	maize, millet, rice, etc.
   Religion:	Christian, Moslem, Hindu, Buddhism, etc.

For examples of scales of measurement, see Annex 8.1. We will come back to these distinctions in Module 22, as continuous, discrete, ordinal and nominal data require different statistical tests.

Factors rephrased as variables

When looking at your problem analysis diagram you will notice that most of what we called ‘factors’ are in fact variables which have negative values. We phrased the contributing factors negatively on purpose (e.g., lack of knowledge) as it is much easier to visualise these factors in the negative. However, in reality not everyone with good knowledge of TB treatment is a regular attender and not everyone with poor knowledge absconds from treatment. As we conduct our study we will try to determine to what extent these contributing factors play a role. Therefore we have to formulate them in a neutral way, so that they can take on positive as well as negative values. The table below presents examples of negatively phrased ‘factors’ and how they can be rephrased as neutral ‘variables’.

Table 8.2: Factors rephrased as variables

Operationalising variables by choosing appropriate indicators

Note that the different values of many of the variables presented up to now can easily be determined. However, for some variables it is sometimes not possible to find meaningful categories unless the variables are made operational with one or more precise INDICATORS. Operationalising variables means that you make them ‘measurable’:

For example:

• In many HSR studies, you want to determine the level of knowledge concerning a specific issue in order to find out to what extent the factor ‘poor knowledge’ influences the problem under study (for example low utilisation of pre-natal care by pregnant women).

  The variable ‘level of knowledge’ cannot be measured as such. You would need to develop a series of questions to assess a woman’s knowledge, for example on pre-natal care and risk factors related to pregnancy. The answers to these questions form an indicator of someone’s knowledge on this issue, which can then be categorised. If 10 questions were asked, you might decide that the knowledge of those with:

  — 0 to 3 correct answers is poor,

  — 4 to 6 correct answers is reasonable, and

  — 7 to 10 correct answers is good.

• Nutritional status of under-5 year olds is another example of a variable that cannot be measured directly and for which you would need to choose appropriate indicators. Widely used indicators for nutritional status include:

  — Weight in relation to age (W/A)

  — Weight in relation to height (W/H)

  — Height in relation to age (H/A)

  — Upper-arm circumference (UAC)

  For the classification of nutritional status, internationally accepted categories already exist, which are based on so-called standard growth curves. For the indicator ‘Weight/Age’, for example, children are:

  — well-nourished if they are above 80% of the standard,

  — moderately malnourished if they are between 60% and 80%, and

  — severely malnourished if they are below 60%.

Defining variables and indicators of variables

To ensure that everyone (the researcher, the data collectors, and eventually, the reader of the research report) understands exactly what has been measured and to ensure that there will be consistency in the measurement, it is necessary to clearly define the variables (and indicators of variables). For example, to define the indicator ‘waiting time’ it is necessary to decide what will be considered the starting point of the ‘waiting period’ e.g., is it when the patient enters the front door, or when he has been registered and obtained his card?

Annex 8.2 gives examples of common variables with different possible choices for indicators.

III. IDENTIFYING INDICATORS IN QUALITATIVE STUDIES

Certain variables cannot be defined with indicators before the study, because the information to do this is lacking. The purpose of the study may be to find this information.

For example, policy makers in Nepal would like to eliminate leprosy. They have noticed that fewer women report for leprosy treatment than men and would like to know whether stigma keeps women from reporting for treatment and/or whether the services have to be more sensitive to the needs of women for privacy at diagnosis.

We define stigma as an undesirable differentness that disqualifies a person from full social acceptance (Goffman: 1963). However, we cannot fill in more precisely in what way men and women are discriminated against, as that has still to be studied. Some indicators for stigma could be the divorce rate of male and female patients, or the degree of isolation of the patient by the healthy spouse or by the community, but how the severity of this isolation should be measured is still unknown. Possibilities includd, for example, whether patients and spouses still share a house, share food, share one bed? Do community members still accept leprosy patients as village leaders, do they welcome patients to attend village meetings, and, if so, do they still drink beer or eat together, and do they ask patients to bring their own cups?

One could state that in exploratory, qualitative studies we study themes, such as stigma, to understand better how patients suffer from stigma and how they cope with it. We also discover contributing factors to stigma: in some societies women are more vulnerable to stigma than men; adolescents are more vulnerable than adults who have settled economically and socially; patients with deformities are always more vulnerable to stigma than those without visible signs.

By better understanding the problem of stigma we can now give an operational definition of the strength of stigma on a scale. This enables us to measure through a quantitative study the degree of stigma male and female patients suffer from, and the most important contributing factors to stigma. (See Figure 8.1)

Figure 8.1: Relationship between qualitative and quantitative studies in understanding and measuring problems

IV. CAUSES AND ASSOCIATIONS; CONFOUNDING

Dependent and independent variables

Because in health systems research you often look for causal explanations, it is important to make a distinction between dependent and independent variables.

For example, in a study of the relationship between smoking and lung cancer, ‘suffering from lung cancer’ (with the values yes, no) would be the dependent variable and ‘smoking’ (varying from not smoking to smoking more than three packets a day) the independent variable.

Whether a variable is dependent or independent is determined by the statement of the problem and the objectives of the study. It is therefore important when designing an analytical study to clearly state which variable is the dependent and which are the independent ones. Note that if a researcher investigates why people smoke, ‘smoking’ is the dependent variable, and ‘pressure from peers to smoke’ could be an independent variable. In the lung cancer study ‘ smoking’ was the independent variable.

Although in everyday language we may speak of possible CAUSES of problems, in scientific language we prefer to speak of ASSOCIATIONS between variables, unless a causal relationship can be proven. If we find an association between smoking and cancer, we can conclude that smoking causes cancer only if we can both demonstrate that the cancer was developed after the patient started smoking and that there are no other factors that could have caused both the cancer and the habit of smoking. Nervous people, for example, may both smoke more and suffer more from cancer than persons who are not nervous.

Therefore, in order to give a true picture of cause and effect, possible confounding variables must be considered, either at planning stage or while doing data analysis.

For example:

A relationship is shown between bottle-feeding and diarrhea in under-twos. However, mother’s education may be related to bottle-feeding as well as to diarrhea.

Mother’s education is therefore a potential confounding variable. In order to give a true picture of the relationship between bottle-feeding and diarrhea of under-twos, the influence of mother’s education should be controlled. This could either be addressed in the research design, e.g., by selecting only mothers with a specific level of education, or it could be taken into account during the analysis of the findings by analysing the relation between bottle-feeding and diarrhea separately for mothers with different levels of education.

Background variables

In almost every study, BACKGROUND VARIABLES, such as age, sex, educational level, socio-economic status, marital status and religion, should be considered. These background variables are often related to a number of independent variables, so that they influence the problem indirectly (hence they are called background variables). Only background variables important to

the study should be measured. Background variables are notorious ‘confounders’.

According to the WHO definition of health as an outcome of the health system (see Module 2, Figure 2.4) we can, for example, measure the improvement in the health of a population by comparing the estimated life expectance at birth and time lived with a disability over the past ten years (provided the epidemiological and other environmental factors did not change). Increased fairness of the health system could be measured by the percentage out-of-pocket spending on health by the poor (living on 1 US$ or less a day) of the total health expenditure, comparing, say, the past ten years.

Responsiveness to patients’ need for human treatment is more difficult to measure, but a number of indicators could be developed, using the concepts: respect for patients (not humiliating or demeaning them); confidentiality with regard to a patient’s diagnosis and treatment, providing patients with essential information, so that they can participate in choices about their own health and treatment, and client-orientedness in the services offered (prompt attention, clean premises) (WHO 2000: 32). It is interesting that one can not only make comparisons within one country over time, but also between countries.

You should continue to adjust your problem analysis diagram, variables and objectives until they are all in line with each other.

REFERENCES

Abramson JH (1990, 4^th ed.) Survey Methods in Community Medicine. London: Churchill-Livingstone. (In particular Chapters 9 and 10)

Moser CA, Kalton G (1979) Survey Methods in Social Investigation. Hants, UK: Gower Publishing Company: 220-224.

World Health Organization (2000) The WORLD HEALTH REPORT 2000. Health Systems: Improving Performance. Geneva: WHO.

Annex 8.1: Example of a framework for defining variables

Annex 8.2: Examples of variables with different options for indicators*

* Adapted from Abramson (1990)

Trainer’s Notes

Module 8: VARIABLES

Introduction and discussion

• Stress that it is important to define the problem as well as the factors influencing the problem in measurable terms.
• Let participants give some examples of numerical variables and discuss what different values these variables may have.
• Let participants give examples of categorical variables, after you have provided one or two examples. Make sure they understand that, once you have clear categories, you can ‘measure’ those variables, which means that you can determine their different values.
• Make sure that the participants understand that certain variables can be ‘measured’ directly and that others need indicators before they can be measured.

  Note: We use quote marks to indicate that ‘measuring’ of categorical variables such as sex or mode of transport means ‘determining their values’.

• Discuss the relationship of the concept of dependent and independent variables to causality and stress that descriptive studies (see Module 4) do not have dependent and independent variables.

Exercise: Examples of dependent and independent variables

• Let the groups give examples from their own studies.
• Explain the difference between association and cause.
• Explain clearly that dependent variables factors with values such as low, medium, and high or sick and well need operational definitions to explain just what these values mean.
• Stress that sometimes measuring variables is not our concern, but rather identifying and describing them (if we know very little about possible causes of a problem).
• Stress the fact that when participants are working to list variables they have identified in their analysis diagram, they should also go back to their objectives to ensure that each objective is adequately covered. Since certain variables may need to be measured for several objectives it would be more complicated to start identifying variables by looking at the objectives rather than the analysis diagram.

Exercise: Identification of variables in research

• Conduct the exercise on ‘Identification of variables in research’ during the plenary session. Ask the participants to read and respond to the questions posed for each problem in the exercise individually or in small groups of two or three people. Give 4-5 minutes for each of the three problems, immediately followed by a group discussion. (Suggested answers are on the following 2 pages.)

Group work

• Ask the participants to meet in their working groups to select the variables that will be involved in the study being designed.
• Each group should then prepare a list of the selected variables for presentation and discussion in plenary and for inclusion in the methodology section of its research proposal. The groups should also indicate which variables would have to be further defined in the field.

ANSWER SHEET FOR EXERCISE 3: Identification of variables in research

(The following answers are by no means exhaustive)

Problem 1:

Problem 2:

Dependent variable:

• Availability of chloroquine for village health workers

Indicator for availability of chloroquine:

‘Short of chloroquine’ should be defined in relation to the time since the date of the last drug supply and ideally also in relation to the size of the population.

For example:

If the number of tablets in stock is measured for all VHWs two weeks after the date of the last meeting at the health centre where drugs were supplied, one could say that any VHW who does not have enough drugs to treat 1% of the population for malaria is short of drugs. Since an adult needs 10 tablets for a full course, this would mean that a VHW should have at least 50 tablets available, if the village has a population of 500.

An alternative definition could be having no tablets in stock two weeks after the last date of supply.

Independent variables:

• Availability of drugs at the Health Centre (influenced by frequency of ordering and frequency of supply)
• Amount of drugs monthly supplied to VHWs
• Number of weeks since the VHW last came in for his supply of chloroquine
• Number of patients treated since date of last supply

Problem 3:

Important independent variables that could be taken into account include:

• Age
• Location in the village
• Contact with water
• Type of farming activities
• Division of labour
• Season

Closer study revealed that schistosomiasis was present among 70% of the young farmers between 20-25 years of age, while it was almost entirely absent in farmers older than 50 years of age. It turned out that younger farmers tended to have farms much further away from the village where the land was more fertile, and they had to cross a river where they bathed on their way home in the evening. The older farmers, on the other hand, had always had their farms close to the village and obtained water from wells.

Document(s) 11 of 27