Differences
This shows you the differences between two versions of the page.
| Both sides previous revision Previous revision Next revision | Previous revision Last revision Both sides next revision | ||
|
8-sampling-strategy [2018/08/01 16:13] thorsten [8.4 Risk-based allocation] |
8-sampling-strategy [2018/08/01 16:13] thorsten [8.1 Sampling at the primary sampling unit (PSU) level] |
||
|---|---|---|---|
| Line 3: | Line 3: | ||
| ===== 8.1 Sampling at the primary sampling unit (PSU) level ===== | ===== 8.1 Sampling at the primary sampling unit (PSU) level ===== | ||
| The information needed to be collected are:\\ A) Number of PSU in the population (which should have been determined in [[/7-Study-design|section 7]])\\ B) Design prevalence\\ C) Desired confidence\\ D) Desired power\\ E) Sensitivity (Sensitivity at the PSU level equals confidence at the SSU level, if two-stage sampling is used)\\ F) Specificity\\ \\ This information will be used in later steps for sample size calculation. In some cases, the sample size is decided a priori, for instance based on a specific budget. If this is the case, then the information in this section would not need to be collected. Then, in step 8.5 (below) you will find a link to tools to explore different sampling options and carry out a power analysis - identify the achievable confidence using the sample size established a priori.\\ \\ | The information needed to be collected are:\\ A) Number of PSU in the population (which should have been determined in [[/7-Study-design|section 7]])\\ B) Design prevalence\\ C) Desired confidence\\ D) Desired power\\ E) Sensitivity (Sensitivity at the PSU level equals confidence at the SSU level, if two-stage sampling is used)\\ F) Specificity\\ \\ This information will be used in later steps for sample size calculation. In some cases, the sample size is decided a priori, for instance based on a specific budget. If this is the case, then the information in this section would not need to be collected. Then, in step 8.5 (below) you will find a link to tools to explore different sampling options and carry out a power analysis - identify the achievable confidence using the sample size established a priori.\\ \\ | ||
| - | ---- | ||
| ===== 8.2 Sampling at the secondary sampling unit (SSU) level ===== | ===== 8.2 Sampling at the secondary sampling unit (SSU) level ===== | ||
| If one-stage sampling process is applied, then PSU is the only unit to consider (skip this step). For two-stage sampling, the same information should be collected for the SSU level:\\ A) Number of SSUs in the population (which should have been determined in [[/7-Study-design|section 7]])\\ B) Design prevalence\\ C) Desired confidence\\ D) Desired power\\ E) Sensitivity\\ F) Specificity\\ \\ | If one-stage sampling process is applied, then PSU is the only unit to consider (skip this step). For two-stage sampling, the same information should be collected for the SSU level:\\ A) Number of SSUs in the population (which should have been determined in [[/7-Study-design|section 7]])\\ B) Design prevalence\\ C) Desired confidence\\ D) Desired power\\ E) Sensitivity\\ F) Specificity\\ \\ | ||
| - | ---- | ||
| ===== 8.3 Selection criteria WITHIN the population ===== | ===== 8.3 Selection criteria WITHIN the population ===== | ||
| In [[/3-Target-population|section 3]] we discussed criteria to choose the target population. When sampling that target population, however, it is possible that not all animals are reachable (for instance not all units are registered, therefore not all are available in the sampling frame). So here a designer should consider whether, WITHIN the target population, there are any criteria for the selection of animal/units to target which are NOT related to sampling (which is discussed further below).\\ \\ Criteria for selection could be for instance logistic/convenience (this component focuses on a particular sector of the population because they are easier to sample or more accessible; higher probability of infection; higher probability of showing clinical signs; feasibility of detection (the diagnostic tests available can only be used in animals above a certain age, or non-vaccinated animals); or higher severity of consequences in case of infection, which is the case for instance when surveillance is focused on breeder animals.\\ \\ | In [[/3-Target-population|section 3]] we discussed criteria to choose the target population. When sampling that target population, however, it is possible that not all animals are reachable (for instance not all units are registered, therefore not all are available in the sampling frame). So here a designer should consider whether, WITHIN the target population, there are any criteria for the selection of animal/units to target which are NOT related to sampling (which is discussed further below).\\ \\ Criteria for selection could be for instance logistic/convenience (this component focuses on a particular sector of the population because they are easier to sample or more accessible; higher probability of infection; higher probability of showing clinical signs; feasibility of detection (the diagnostic tests available can only be used in animals above a certain age, or non-vaccinated animals); or higher severity of consequences in case of infection, which is the case for instance when surveillance is focused on breeder animals.\\ \\ | ||
| - | ---- | ||
| ===== 8.4 Risk-based allocation ===== | ===== 8.4 Risk-based allocation ===== | ||
| Before establishing the sampling method and details, a surveillance designer should consider the risk characteristics identified in [[/1-Surveillance-system|section 1.5]]. In the RISKSUR framework these risk characteristics are presented back to the user at this stage. If risk-based sampling can be applied to the identified risk strata, this step should be used to define the strata and collect information needed for sample size calculation and allocation. Risk based strategies have been applied to improve the efficiency of surveillance for early detection (particularly for vector borne disease or those with a wildlife reservoir), case detection and demonstrating freedom. It is important when using risk based strategies particularly when doing surveillance for early detection that the possibility of disease occurrence outside the high risk population is considered and additional surveillance (e.g. passive surveillance) is carried out to detect these cases if required.\\ \\ The link below provides tools that can be used for risk assessment, which would use the information on the distribution of the hazard and the risk in the population in order to determine risk strata for sampling, and estimate some of the information to be recorded for each strata (detailed below).\\ \\ {{:files:tools.jpg?nolink&50 |}}__**RISK ASSESSMENT TOOLS (LINK)**__.\\ \\ For each risk stratum defined, the following information should be recorded:\\ A) Defined due to a higher risk of: Infection, Detection, Consequences, or others? (visit [[/1-Surveillance-system|section 1.5]] or the [[/Glossary|Glossary ]]to understand the differences)\\ B) Any further description - p.e. if based on geographical factors, temporal, etc\\ C) Percentage of the population\\ D) Risk ratio\\ \\ | Before establishing the sampling method and details, a surveillance designer should consider the risk characteristics identified in [[/1-Surveillance-system|section 1.5]]. In the RISKSUR framework these risk characteristics are presented back to the user at this stage. If risk-based sampling can be applied to the identified risk strata, this step should be used to define the strata and collect information needed for sample size calculation and allocation. Risk based strategies have been applied to improve the efficiency of surveillance for early detection (particularly for vector borne disease or those with a wildlife reservoir), case detection and demonstrating freedom. It is important when using risk based strategies particularly when doing surveillance for early detection that the possibility of disease occurrence outside the high risk population is considered and additional surveillance (e.g. passive surveillance) is carried out to detect these cases if required.\\ \\ The link below provides tools that can be used for risk assessment, which would use the information on the distribution of the hazard and the risk in the population in order to determine risk strata for sampling, and estimate some of the information to be recorded for each strata (detailed below).\\ \\ {{:files:tools.jpg?nolink&50 |}}__**RISK ASSESSMENT TOOLS (LINK)**__.\\ \\ For each risk stratum defined, the following information should be recorded:\\ A) Defined due to a higher risk of: Infection, Detection, Consequences, or others? (visit [[/1-Surveillance-system|section 1.5]] or the [[/Glossary|Glossary ]]to understand the differences)\\ B) Any further description - p.e. if based on geographical factors, temporal, etc\\ C) Percentage of the population\\ D) Risk ratio\\ \\ | ||