Identifying and analyzing variables in musculoskeletal healthcare research

The causal relationship

In studies investigating associations, four sets of variables may be involved:

• Change variables instigate the alterations in a phenomenon, situation, or circumstance

• Outcome variables (OVs) signify the effects, impacts, or repercussions of a change variable

• Variables influencing the connection between cause-and-effect variables

• Connecting/linking variables play a crucial role in finalizing the relationship between the “cause-and-effect” variables in any given scenario.

IV And DV

The IV is the predictor variables. They can also influence other variables.^[1] IV represents any condition, intervention, or characteristic capable of predicting or causing a specific outcome. Conversely, the OV, also known as the DV, is defined as a variable whose value depends on or is a consequence of another variable. Therefore, the variable preceding the DV is termed the IV.^{[1-3,7,10,12]}

The operational definition (OD) should be sufficiently detailed so other researchers can replicate the condition/procedure. IV is operationally defined as operationalized accordingly, how they are manipulated by the researcher/investigator.^[1,3] For example, a therapist prescribes a group of resisted exercises to gain strength in the biceps muscle. In this scenario, the IV will be the resisted exercises, which can be manipulated to observe their effects on the DV, i.e., strength gain. The DV (strength gain) is the outcome that the researcher measures to determine if the IV has an effect.

OD is crucial in research as it provides clear and specific guidelines for measuring or assessing variables.^[1,2,12] For example:

• Variable: LBP

• OD: Numeric pain rating scale (NPRS).

NPRS is commonly used to assess the intensity of pain. It typically consists of a scale from 0 to 10, where 0 represents no pain and 10 represents the worst pain. This OD ensures that researchers and readers clearly understand how LBP is assessed and measured in the study, facilitating transparency and reproducibility in research.

Moderating variable

The variable has enough strong conditional influence to modify the original relationship between the IV and DV.^[7] It moderates the strength and the relationship between the dependent and IV. For instance, a study into the effect of quality of instruction on players’ sports may have moderating variables like the player’s interest. This implies that the students interested in a particular activity will likely perform better in-ground performance. In contrast, the other students without such an interest may not perform well, irrespective of how good the instructions are.

Intervening variable

Intervening variables are sometimes also known as confounding variables (CVs).^[1] In definite conditions, the connection between DV and IV is contingent upon the presence of another variable. Only in the existence or occurrence of an intervening variable can the IV, or the cause, express itself.^[1-3]

Active variables (AVs)

AVs are those influenced by the researcher/experimenter assigning subjects to IV levels.^[1-3,5,6] For example, in any study where participants are assigned to receive the medication and the placebo, the treatment is an AV.

Attribute variable

The researcher cannot allocate subjects to groups; rather, they have to observe participants within their accepted groupings based on inherent characteristics. Attribute variables cannot be manipulated by the experimenter.^[1] When studying the effects of one or more attribute variables, a research endeavor cannot be classified as a true experiment.^[1,13] For instance, occupation serves as an attribute factor wherein subjects naturally belong to specific groups without the possibility of assignment.^[1] Other pre-existing characteristics such as age, gender, and diagnosis also fall under attribute variables.^[1,13] For instance, examining differences in posture and range of motion (ROM) across various age groups and between genders such as males and females highlights this. Since sex is an inherent attribute before the study, relationships can be observed, but the design lacks adequate control, thereby restricting the interpretation of cause and effect.

It is achievable to combine the attribute variable and the AV in a one or single study.^[1,13] As an illustration, one could examine the impact of aerobic exercises on Type II diabetes mellitus alongside the influence of age. This could involve categorizing participants into four age groups, with each individual within a group randomly assigned to one of two intervention groups. Despite involving an attribute variable, this study qualifies as an experimental because the researcher has the freedom to manipulate the allocation of intervention levels for at least one IV.

Quantitative variable

Quantitative variables reflect the notion of magnitude. They represent the numerical (numbers) and the measurements,^[14,15] e.g., when the therapist measures the shoulder’s ROM with the goniometer and assigns a numerical value representing the degree of movement (quantitative variable). The quantitative variables are divided into two types: Discrete and continuous.

Continuous variables

They can assume any theoretical value within a defined numeric range.^[5] An indefinite number of fractional values exists between any two continuous variable values. In reality, the accurate measurement of continuous variables is unattainable, constrained solely by the precision of the measuring instrument.^[1,5,15] For instance, the ROM of a joint can be recorded as 50°, 50.4°, or even 50.2°, contingent upon the marking or gradation of the goniometer and the proficiency of the examiner.

Interval variables

Interval variables can be measured along a continuum in a scale. The feature/factor of rank order can be found in interval variables,^[6] e.g., temperature (which could be measured in Celsius and Fahrenheit). The scale or outcome measure is not measured from the zero, instead it is measured from the normal established value. Like the normal body temperature is defined as around 98.6°F, and more than that will be accounted as pyrexia. A zero-degree Celsius (0°C) reading does not indicate a complete lack of temperature.

Discrete variable

If variables can only be shown as numbers and they are non-continuous (don’t take all possible values in a range), they are called discrete variables. Discrete variables have specific, separate values, like whole numbers.^[16] Discrete variables are those that can only be complete, whole units.^[1,16] As an illustration, heart rate is always quantified in beats per minute rather than in fractions of beats. Qualitative variables also describe discrete categories, such as sex (male/female).

Qualitative and categorical variable

In several ways, categorical variables are similar to qualitative variables. Both variables use either an ordinal or nominal measurement scale. Still, some variances are observed.^[17]

For instance, it is possible to evolve classifications based on measurements derived from a continuous scale.

Overall, there is a nuance between categorical and qualitative and continuous and quantitative variables.^[5] Categorical variables are examined/measured on ordinal/nominal measurement scales whereas the continuous variables are measured on an interval or ratio scale.^[13]

The categorical variable is divided into three categories:^[13,16]

1. Constant variable: These variables have only one category or a single value. For example, the therapist repeatedly practices the same task under the same condition.

2. Dichotomous variables are when qualitative variables have only two categories or two levels. They may have similar features to nominal variables in terms of not having any intrinsic variable, such as sex (male/female).

3. Polychotomous variables: These variables have three or more values or categories of subsets, e.g., built (mesomorph, ectomorph, and endomorphic).

RD and variables

The RD can be configured to enable comparison between the independent groups of participants, or it may involve comparing responses across the treatment conditions within each participant.^[1] When the IV level is assigned to numerous groups with distinguishing attributes or AVs, the IV is regarded as an independent factor.^[1,2,5] For instance, in a study comparing the impact of two types of mobilization on pain reduction and increasing ROM in patients with bilateral periarthritis, each type would be administered to distinct patient groups.^[18] Hence, the variable “mobilization” constitutes an independent factor with two levels.^[19] Similarly, if the study assesses the effects of mobilization between male and female participants, “sex” would also function as an independent factor.

In exploratory studies, the IV and DV are typically assessed simultaneously to determine if they exhibit a predictive correlation.^[1] For instance, researchers have examined the connection between LBP and various factors such as age, sex, cognitive status, ambulatory status, analgesic use, osteoporosis, and osteoarthritis within a long-term care population. The DV, representing the outcome, was the presence of LBP, while the IV acted as predictor variables, including characteristics such as age, sex, and cognitive status. Such studies often encompass multiple IVs as researchers seek to discern the interconnectedness of different factors in explaining the OV.

Experimental studies entail comparing various conditions to explore causal relationships, wherein the IV is manipulated while the DV is measured.^[1,2,20]

For example, researchers compared the effectiveness of core strengthening and usual medical care to assess if core strengthening was effective in relieving pain in patients with LBP. The study measured outcomes such as changes in disability scores and the pain scale ratings. In this scenario, the IV is core strengthening (intervention), while the two DVs are disability and pain scores (response). Any alteration in the DV is assumed to be influenced by the condition of the IV; in other words, the DV is contingent upon the state of the IV.

In comparative studies, the RD involves considering multiple IVs.^[2] For instance, researchers may examine the patient’s sex in addition to the intervention to determine whether the effectiveness of core strengthening varies between the sexes (males and females). In this case, there are two IVs: Type of intervention and sex. Furthermore, a study can encompass more than one DV.

Demographic variable

The variables are collected by the researcher to elucidate the nature and the distribution of the sample employed alongside inferential statistics. The variables such as age, sex, occupation, ethnicity, and socioeconomic measure. Descriptive statistics is often used in demographic variables. It may also enter into multivariate models for confounding and controlling effects.^[21]

Levels of IV

In comparative studies, IV is described as “values” referred to as levels.^[1] In research, the levels represent the groups of conditions under comparison. Each IV typically comprises a minimum of two levels. Unlike IV, DVs are not categorized into levels.^[1] For instance, in the study comparing spinal extension and home care, the IV “intervention” has two distinct levels: Spinal extension and home care. If the study had incorporated additional interventions such as spinal manipulation, mobilization, or bed rest, it could alter the intervention variable’s level count rather than the number of variables themselves.

Manipulation of the variables

Manipulating the variables involves intentionally setting specific experimental conditions (the IV) for at least one group of participants/subjects. This deliberate action is carried out by the researcher or experimenter. This process allows the researcher to control and study the effects of the manipulated variables on the DV.^[1] The experimenter adjusts the IV levels by assigning subjects to different conditions. This typically involves providing the intervention to one group while refraining from doing so with another.^[1,22] For instance, the researcher may wish to investigate the impact of medication on reducing hypertension. Subjects can be assigned to treatment and control groups, and their blood pressure changes can be measured after a period of receiving treatment or no treatment. It is feasible to manipulate either a single variable or multiple variables concurrently.

EV

Factors not directly pertinent to the study’s objectives but capable of influencing the DV are termed EVs.^[14,17,23] It is also called “nuisance variables” or intervening variables.^[5] EV could be external factors stemming from the environment and experimental conditions, as well as internal factors reflecting the personal characteristics of the study’s subjects or participants.^[1] For example, in a study, researchers are examining the effect of a new manual protocol for LBP. However, the EV can be the use of painkiller medication, mood, tiredness, or the subject’s behaviors. The EV could affect the DV (LBP) [Figure 2].

Close

Figure 2:

Extraneous variable could potentially affect the dependent variable.

Export to PPT

CV

CVs are often referred to as nuisance variables.^[2] The confounding occurs when EVs disrupt the observed relationship between an exposure and an outcome.^{[1,2,9,22,24]}

• CV correlates with exposure

• CV poses a risk for the disease irrespective of the exposure

• CV does not form a direct part of the causal pathway between exposure and the disease. To clarify, a CV is linked to the predictor variable but may also contribute to the risk of the OV, necessitating its consideration. Confounding arises when the exposure is potentially influenced or distorted by EVs.

For example, in a study evaluating the effectiveness of a certain exercise program for knee pain, the age of the participants could act as a CV. This is because age can independently affect knee pain and also influence the response to the exercise program. Therefore, the age of the participants needs to be controlled for or stratified in the analysis to ensure that its influence on the outcome is not mistaken for the effect of the exercise program.