Survey of the Medical Devices Industry

Detailed information for fiscal year 2000

This survey produces statistical information on firms engaged in medical devices production in Canada.

Data release - December 5, 2002

• Questionnaire(s) and reporting guide(s)
• Description
• Data sources and methodology
• Data accuracy

Description

The survey of the medical devices production industry produces statistical information on diverse aspects of the industry: employment, revenues, goods manufactured, investments, research and development. The data collected from this survey will be used to determine the competitive position of the sector, which will aid in the development of trade and investment promotional materials, and to determine the need for present and future federal government policies and programs for the medical devices industry.

The survey is being conducted on behalf of Industry Canada's Life Sciences Branch and Health Canada's Therapeutic Products Directorate. The survey is endorsed by Medical Devices Canada (MEDEC); the Association québécoise des fabricants de l'industrie médicale (AQFIM); the Association of Ontario Medical Manufacturers (AOMM); the Health Care Products Association of Manitoba (HCPAM); the Metro Edmonton Health Industry Association (MEHIA); the Calgary Association for Medical Products (CAMP); the BC Medical Device Industry Association (BCMeDIA); the Nova Scotia Biotechnology and Life Sciences Industry Association (BioNova) and the Ottawa Life Sciences Council (OLSC).

Subjects

• Manufacturing
• Other manufactured products

Data sources and methodology

Instrument design

The content of the questionnaire is primarily developed by Industry Canada and Health Canada. Statistics Canada formatted the questionnaire and collaborated with Industry Canada and Health Canada on developing the content.

Sampling

This is a sample survey.

Industry Canada and Health Canada provide the survey frame that is used for this survey. The frame contains 755 unique business units. Statistics Canada conducts a census of the units on the survey frame.

The following table shows the provincial distribution of the business units:

Newfoundland, 7 units
Prince Edward Island, 7 units
Nova Scotia, 19 units
New Brunswick, 16 units
Québec, 144 units
Ontario, 406 units
Manitoba, 31 units
Saskatchewan, 10 units
Alberta, 46 units
British Columbia, 69 units
Canada, 755 units

Data sources

Data collection for this reference period: October 12, 2001 to January 18, 2002

Responding to this survey is voluntary.

Data are collected directly from survey respondents.

The questionnaire is a mail out/mail back survey. Respondents are given the choice to complete the questionnaire and send it back by mail or fax, or else complete the questionnaire over the phone with a representative from Statistics Canada. Follow-up calls to non-respondents start a month after the mail out, and continue for two months.

View the Questionnaire(s) and reporting guide(s) .

Error detection

A number of consistency edits are applied to the data. Consistency edits ensure the reported total net sales and revenues from medical devices are equal to the sum of the reported breakdowns, and that all percentage breakdowns equal 100%.

Imputation

The imputation is performed using the nearest donor method. This method involves determining an appropriate donor using a set of agreement and similarity variables. The potential donors and the recipient must have exactly the same values for the agreement variables. For example, a company in a specific activity (e.g. manufacturing only) is only a potential selected donor or recipient for that specific activity. With respect to the similarity variables, the object is to minimize the relative distance between each potential donor and the recipient for the selected criterion. Examples of similarity variables include revenue from a medical device and total employment assigned to a medical device. For the variables being imputed, the values missing from the recipient is filled in with the corresponding values from the selected donor.

Estimation

Even though the survey is designed as a census survey, because not all of the population is observed, values for the figures of interest have to be estimated; a stratified single-stage sample design is used. The strata are based on type of firm (firms identified by survey sponsors as "important" and "other" firms) and region (Atlantic, Quebec, Ontario, Manitoba, Saskatchewan, Alberta and British Columbia), for a total of 14 possible strata. To compensate for total non-response, the units are reweighted by stratum (ratio of the number of companies in the population to the number of companies in the sample).

Parameters of interest are estimated with Statistics Canada's Generalized Estimation System (GES). The Generalized Estimation System automatically produces estimates for each variable by area of interest (e.g., firm size, sector) and a measure of precision (coefficient of variation (CV)).

Quality evaluation

Existing data sources from Statistics Canada and Industry Canada are examined, but because the survey is an activity based survey, not industry based, it is difficult to find comparable information.

Disclosure control

Statistics Canada is prohibited by law from releasing any information it collects which could identify any person, business, or organization, unless consent has been given by the respondent or as permitted by the Statistics Act. Various confidentiality rules are applied to all data that are released or published to prevent the publication or disclosure of any information deemed confidential. If necessary, data are suppressed to prevent direct or residual disclosure of identifiable data.

Confidentiality analysis includes the detection of possible direct disclosure, which occurs when the value in a tabulation cell is composed of a few respondents or when the cell is dominated by a few companies.

Data accuracy

Because the figures of interest are estimated, the survey has some sampling error, expressed as a CV. The CV is a percentage that expresses the size of the standard error as a proportion of the estimate to which it is related. For example, a CV of 10% indicates that the standard error is 10% of the estimate. If the annual revenue estimate is $9.0 million, with a CV of 10%, then the standard error is $0.9 million. CVs are used to rate the quality of each estimate, using the following table:


Value of CV Code Rating

0 to 5% A Very Good
5.01% to 15% B Good
15.01% to 33% C Good to Poor-- use with caution
33.01% and over D Very poor -- may not be acceptable

Most estimates are rated either good or very good.

Survey estimates may also contain non-sampling error. Non-sampling errors are not related to sampling and may occur for many reasons. Population coverage errors, differences in the interpretation of questions, incorrect information from respondents, and mistakes in recording, coding and processing data are examples of non-sampling errors. Non-response is an important source of non-sampling error. While the impact of non-sampling errors is difficult to evaluate, measures such as response rates and imputation rates can be used as indicators of the potential level of non-sampling error.

The response rate in the Survey of the Medical Devices Industry was 55.5%.