Most important terms related with Health Technology Assessment.

the absolute arithmetic difference between the risk of an negative endpoint in the control group and the risk in the experimental group. ARI determines how much the risk of an adverse endpoint increased as a result of the intervention.

the absolute arithmetic difference between the risk of an adverse endpoint in the control group and this risk in the experimental group. ARR determines the absolute value of risk reduction. The greater the ARR, the greater the intervention’s impact. In case two interventions have identical impact, ARR is zero.

an absolute parameter used for assessing positive endpoints, calculated as the difference between the probability of an event in the experimental group and the control group. Values greater than 0 mean that the probability of a positive endpoint in the experimental group is higher than in the control group, thus BD> 0 evidences the advantage of the experimental group.

Inference about the statistical significance of the benefit difference – if the confidence interval for the BD parameter does not amount to 0, then the difference is statistically significant, while if it contains 0, then the difference is not statistically significant.

is the simplest modelling method. The essence of this technique is determined by its name – by means of nodes and branches coming out of them, we can draw paths for patient management. There are three types of nodes:

- decision node – further action depends on a decision (e.g. a physician’s decision on what second-line treatment to apply after first-line treatment failed or a patient’s decision to accept or refuse surgical treatment),
- chance node – further action depends on likelihood of specific events (e.g. whether as a result of the therapy applied the patient survives or dies),
- end node – terminates the pathway (e.g. recovery or death).

Flexibility characterising decision trees makes it possible to add or eliminate branches depending on how accurately the reality has been represented (excessive simplification or complexity of a problem). A decision tree may constitute a basis for further simulations, e.g. in Markov models.

is an organised activity of a particular system of health care based on health protection institutions (Mała Encyklopedia Medycyny PWN, PWN Concise Encyclopedia of Medicine).

is an organised activity aimed at disease prevention and treatment; activities for the health of citizens, which is included in the system corresponding to the systemic assumptions of State (Mała Encyklopedia Medycyny PWN, PWN Concise Encyclopedia of Medicine). Health protection includes health care and activities of other divisions affecting the health status of the population (construction, food industry, municipal economy, water, sport, tourism and others). The concept of health protection is therefore much broader than health care.

is the cost of obtaining one unit of additional health effect, as a result of substituting the standard treatment (comparator) with the analysed intervention. Due to limited funds in the health sector, this ratio is very helpful in making decisions on whether to finance a given medical procedure, such as screening or a drug.

Inference about the statistical significance of the benefit difference – if the confidence interval for the BD parameter does not amount to 0, then the difference is statistically significant, while if it contains 0, then the difference is not statistically significant.

the difference between mean values in the experimental group and the control group. Values lower than 0 indicate that the value of a given parameter in the experimental group is smaller than the value of that parameter in the control group. In the case of negative endpoints, MD

is a mathematical modelling technique based on matrix calculus. They describe the transition of patient cohorts between mutually exclusive and exhaustive health conditions during a series of consecutive cycles (time intervals between consecutive transitions between the states). The model allows for taking into account the disease’s development over time. It is particularly useful in modelling chronic diseases.

Markov models are characterised by a number of possible states (e.g. Well, Diseased, Dead) and rules of transition between them.

is the number of patients in whom a given intervention leads to occurrence of one additional adverse endpoint in a given time period. It is calculated as the inverse of the absolute risk increase (ARI).

means the number of patients who need to be subjected to an intervention in order to obtain the desired health effect or avoid one adverse endpoint in one patient in a given time period. Calculated as the inverse of absolute risk reduction (ARR).

refers to four elements: population, intervention, comparison, outcome.

- P – population — a finite group which the researcher wishes to investigate.
- I – intervention — a drug, therapeutic procedure, diagnostic test, time.
- C – comparator (also: alternative technology, compared technology) — a health technology (intervention) compared to the technology under consideration.
- O – outcome — parameter used for assessing and measuring effects of the intervention.

is an experimental study on assessment of efficacy of a health technology in which patients are randomly assigned to the experimental group and control group and subsequently results in both groups are compared. The experimental group receives the intervention which is the subject of the study and the control group receives the standard intervention or placebo. Randomised trials are considered to be the most reliable and particularly useful for assessing efficacy and safety of preventive and curative interventions.

is a parameter used to assess negative endpoints, such as the occurrence of death, stroke, recurrent AF or side effects. RR specifies how many times the probability of a given endpoint has increased (or decreased) after an intervention was applied (as compared to the control intervention). RR values greater than 1 imply increased likelihood of adverse events in the experimental group, thus RR > 1 should be construed as advantages of the control intervention, and RR

Inference about the statistical significance of the relative risk – if the confidence interval for the RR parameter does not amount to 1, then the difference is statistically significant, while if it amounts to 1, then the difference is not statistically significant.

is a form of data research conducted in line with predefined inclusion and exclusion criteria, e.g. for clinical trials, regardless of the results of individual studies. Qualitative review of scientific reports conducted with the aim of solving a research problem can be considered systematic research provided it is characterized by:

- a formal and full process of searching for credible scientific reports,
- clear setting of objective a priori inclusion and exclusion criteria for clinical trials,
- a quantitative statistical analysis of study results, including possible results compiled in meta-analyses.

A review can be considered systematic provided at least 4 out of 5 criteria are met:

- the research question is clearly determined,
- the search strategy is presented,
- inclusion and exclusion criteria for primary clinical trials are predetermined,
- a critical appraisal of included clinical trials, evaluation of their strong and weak points in terms of methodology is carried out,
- a statistical analysis involving a compilation of homogeneous clinical studies is properly carried out and conclusions are properly drawn. [Reference to Cook’s criteria]

is a parameter evaluating the joint effect of a given intervention on the basis of individual studies in which the effect was measured, using the same scale. When calculating WMD results of individual studies are assigned weights depending on their precision. Similarly as in the case of MD, values lower than 0 indicate that the value of a given parameter in the experimental group is smaller than the value of that parameter in the control group. In the case of negative endpoints, MD < 0 evidences the advantage of the intervention which is the subject of the study. In the case of positive endpoints, MD < 0 evidences the advantage of the control group.

Inference about the statistical significance of the mean difference – if the confidence interval for the WMD or MD parameter does not include 0, then the difference is statistically significant, while if it includes 0, then the difference is not statistically significant.