Name: Lot Quality Assurance Sampling Survey.
Origin: Lot Quality Assurance Sampling (LQAS) Survey was originally developed and used in the manufacturing sector for product quality control. In order to minimize the cost of quality control, a sample of "lots" of goods are inspected; those lots with less than a pre-determined number of defective pieces are considered of acceptable quality, and those lots with more than the number of defective pieces are "rejected".
Basic Description: The surveyed population is to be divided either geographically or administratively into smaller "lots". In each lot the target survey "unit" should be identified; for example, "a fully vaccinated child". Those lots with less than a pre-determined allowable number of defective units, for example a child who is "not fully vaccinated", are considered of acceptable quality, and those lots with more than the allowable number of defective units are "rejected". A sample size in each "lot" is determined by the level of precision desired. The sample size and the maximum allowable "defective units" per lot are provided in probability standard tables. For example, a precision of 80% requires a sample of 13 units per lot and an acceptable lot can not have more than four defective units.
Several features of this survey method have made it advantageous for use in public health program monitoring. Through a LQAS survey, reliable information can be obtained on a small geographic or administrative unit using a small sample. LQAS can be used to accurately detect the extremes of performance; those which are exceeding an "upper threshold" of performance and those fail to meet a "lower threshold" of performance. LQAS cannot detect performance levels between those arbitrarily set upper and lower thresholds.
Country Applications:
| Country | Year | Purpose |
| Bangladesh | 1996 | Immunization coverage in urban areas |
| Burkina Faso | 1994 | Immunization Coverage |
| China | 1994 | Immunization Coverage |
| Costa Rica | 1988- 90 | Immunization Coverage & Injection Safety |
| Congo | 1988 | Immunization Coverage |
| India | 1992 | Immunization Coverage |
| Indonesia | 1986 & 1996 | Immunization Coverage & Neonatal Tetanus Mortality |
| Kenya | 1995 | Contraception Coverage |
| Malawi | 1992 | Check records of village health workers for completion |
| Mozambique | 1993 | Immunization Coverage, diarrhea prevalence |
| Peru | 1984-1988 | Immunization Coverage, ORT |
| Senegal | 1996 | Contraceptive use |
| Turkey | 1996 | Immunization Coverage |
| Uganda | 1996 | Trypanosomiasis serosurvey |
Modified from: Robertson S. et al, 1997, The Lot Quality Technique: a global review of applications in the assessment of health services and disease surveillance. World Health Statistics Quarterly, No. 50.
Languages: English, French, Bangla
Technical Scope: Service coverage and quality assurance
Purpose: To assess selected health service coverage or quality in a particular geographic or administrative area, using a small sample size.
Type of methods: Quantitative
Key users information
LQAS is meant to assist local managers to monitor the performance or the coverage of health services in their catchment areas. The survey points out to managers areas with obviously low service coverage and areas with obviously high service coverage.
Objectives and Scope of Tool
Due to resource limitations, managers are interested in finding out where supervision should be focused. Instead of spreading scarce supervision resources equally to all catchment area, LQAS enables managers to identify low performing areas according to the "upper threshold" and the "lower threshold" of performances specified before the survey. Lots which perform above the upper threshold are obviously acceptable and need little attention. Lots which perform lower than the lower threshold are "rejected" and need focused attention.
Key Monitoring Needs and Evaluation Questions Tool Seeks to Address
LQAS is meant to identify high and low performing areas such as high and low vaccination coverage areas, high and low oral rehydration salt use areas or high and low contraceptive use areas. The evaluation question the tool is seeking to answer is: In a specific catchment area, which sub sections of the catchment area are clearly performing well and which ones are clearly failing to meet the basic achievement standards? The tool can not identify those areas that are performing between the two extremes.
Key Indicators
As LQAS is adapted from the manufacturing industry, its use in public health is still limited. So far, this has mostly been employed to monitor immunization coverage. In this case, key indicators are the same indicators employed for EPI programs e.g. measles coverage ratio for children 12 months old, ratio of BCG, Polio, DPT vaccination for children 12 months old, ratio of mothers of children under 12 months old who are vaccinated twice by tetanus toxoid (TT2). There has been limited use of the method in other public health areas such as monitoring ORT use or contraception use. It has also been used to measure serosurveys e.g. Uganda Trypanosomiasis serosurvey.
Research Design
Like other sampling approaches developed for application in industry, LQAS is oriented toward practical action. In primary health care, managers at the local level have few tools available for determining the extent of service coverage. Due to resource limitation, any realistic strategy for collecting information on health services coverage must carefully avoid excess precision. LQAS offers this attribute by identifying areas to focus scarce supervisory resources.
The manager must choose the level of staff performance that defines adequate quality (upper threshold), and the minimum level requiring managerial response (lower threshold). For example, the manager may select 80% polio vaccination coverage as a standard of adequacy, and choose 50% as the performance level at which managerial attention is a priority.
In a health service with performance standards such as immunization coverage, areas which have reached or exceeded the upper performance threshold, for example 80%, are precisely identified so that resources are not unnecessarily invested in them. Conversely, health workers who reach or fall below the lower threshold, for example 50%, are identified so that attention can be given to these priority areas. LQAS can not detect performance between the upper and lower thresholds. However, this limitation is not particularly severe. This level of decision making is exactly what field managers require to focus their limited resources on upgrading performance in the obviously weak coverage areas, and avoid wasting time in supervising obviously strong coverage areas. (Valadez, 1991, Assessing Child Survival Programs in Developing Countries)
Definition of a "Lot"
The surveyed catchment areas need to be divided into subareas "lots". This division could use geographic boundaries, e.g. villages or slums, or administrative boundaries, e.g. Zones, Neighborhoods or Wards. In Bangladesh, for example, measuring immunization coverage in urban areas required the division of target cities by its administrative Wards.
Calculation of Sample Size, and Accept/Reject Levels
Before the survey, managers need to decide what is the "upper threshold" of performance and what is the "lower threshold". For example, in Bangladesh, the national immunization coverage goal of "85% fully immunized by one year" was selected as the upper threshold. Lots which pass the upper threshold are "accepted". The lower threshold was identified based on results of previous coverage surveys and was set at 60%. Lots with coverage below this would be considered to have unacceptable level of performance and hence "rejected".
Level of Precision
Before the survey, managers also need to decide on the level of precision of their survey results. The higher the required level of precision the larger the sample size and the cost of the study. For example, an 80% level of precision (there is 20% probability of mis-classifying accepted lots as rejected ones or vice versa) requires a lot sample size of thirteen, and the maximum allowable number (cut off point) of defective units (e.g. partially immunized children) is four.
Definition defective units
The units to be examined in each lot needs to be identified. For example, in measuring immunization coverage, children aged 12 to 23 months are the survey units. A child is "defective" if not fully vaccinated by the age of 12 months.
Sample Selection
The required number of units in each lot needs to be randomly selected. This requires either pre-registration of all households and children in the survey catchment area, or use of previous registration records.
Training
Particular attention should be given to ensuring that the survey team understands the definition of survey "lots" and survey "units". The survey instrument must be designed to ensure that"defective units" will be identified by the questions. For example, in measuring immunization coverage, the survey should investigate, for all children included in the sample, each child's date of birth and date of vaccination for each antigen received. Otherwise, the analysis will not be able to reveal which child (unit) is defective (not fully vaccinated by 12 months of age).
Lessons from experience:
Before using LQAS, it is important to understand what it can and can not do. It is a method that
uses a small sample size to detect with precision the extreme ends of health performance i.e.
obviously good performance and obviously poor performance. It is not intended to detect the
exact level of performance.
It is important to understand the methodology before collecting data. This is because the sample size depends on the decided upper and lower thresholds as well as the level of desired precision. Principal investigators should have a basic understanding of the LQAS methodology. Trying to explain the methodology to field workers can complicated and is not always necessary. Field workers need to be told about the survey unit, the sample size/lot and how to select the units in each lot.
While LQAS is mainly designed to inform managers whether a certain lot has or has not achieved a certain level of performance, aggregation of observed units in all the lots may give an ideas about the level of overall performance. For example, studying immunization coverage in ten lots by selecting 19 children in each lot will result in a total of 190 observations on immunization coverage in the ten lots combined. If such results are weighted by the size of population in each lot then overall immunization coverage can be calculated.
Analysis
Experience with analysis of the LQAS survey showed that it is easy and does not require a sophisticated statistical package. Due to the small sample size in each lot, the analysis can be done using a simple spreadsheet.
Manuals and Guidelines
Lemeshow S. et al, 1991, Lot Quality Assurance Sampling: Single-and Double Sampling Plans. World Health Statistics Quarterly 44: 115-132.
Valadez J., 1991, Assessing Child Survival Programs in Developing Countries - Testing Lot Quality Assurance Sampling. Department of Population and International Health, Harvard School of Public Health. Harvard University Press.
World Health Organization, 1996, Monitoring Immunization Services Using the Lot Quality Technique. WHO/VRO/TRAM/96.01.
Example of LQAS Questionnaire
The attached is an example of the LQAS questionnaire used in Bangladesh to measure immunization coverage.
Contact Person
Youssef Tawfik
BASICS Project
1600 Wilson Blvd.
Arlington, VA 22209
703-312-6800