Volume 8: No. 1, January 2011
TOOLS AND TECHNIQUES
|Box. Questionnaire Topics, Italian Behavioral Risk Factor Surveillance System (PASSI), 2007-2008|
|Fixed core modules|
|Self-perceived health and quality of life||All|
|Cervical cancer screening||Women, 25-69 y|
|Breast cancer screening||Women, 40-69 y|
|Colorectal cancer screening||Women and men, 50-69 y|
|Diet and nutritional status||All|
|Cardiovascular risk factors||All|
|Rubella vaccination||Women, 18-49 y|
|Prevention of traffic accidents||All|
|Prevention of domestic injuries||All|
|Regional optional modules (since June 2008)|
|Prevention of traffic accidents (supplement)||All|
|Hormone replacement therapy||Women, 45-60 y|
|Child health promotion campaign||All|
|Abbreviation: PASSI, Progressi Delle Aziende Sanitarie per la Salute in Italia (Progress by Local Health Units Towards a Healthier Italy).|
The questionnaire covers a variety of topics related to health behavior and
prevention, all of which are considered priorities in the National Health Plan (www.ministerosalute.it/dettaglio/
phPrimoPiano.jsp?id=316) and the National Prevention Plan (www.ccm-network.it/ en_National_Prev_Plan) (Box). Particular attention is given to subjective aspects, such as the respondents’ perceptions, opinions, knowledge, and attitudes about health behaviors, and whether their doctors provide them with appropriate medical advice. Almost all questions are closed-ended, with multiple-choice answers. Many questions are administered only to specific population subgroups.
The questionnaire has 4 parts:
The questionnaire is revised annually so that new questions can be added (or existing ones reworded), based on public health policy and emerging evidence on the issues addressed by the surveillance system. In the first 2 years of data collection (2007-2008), the fixed core had 2 minor revisions, and 3 optional modules were added. In late 2009, an emerging module on the A-H1N1 influenza virus was implemented, and in 2010 a revised version of the questionnaire was introduced that included a rotating module and 2 new optional modules.
In each LHU, a random sample is extracted each month from the enrollment lists of residents aged 18 to 69 years in the catchment area, stratified by sex and age group (18-34 y, 35-49 y, 50-69 y). These lists contain demographic data (eg, birthday, sex), home address, name of their general practitioner (GP), and often a telephone number. The sample is large enough (at least 25 people per month per LHU) so that annual estimates of the main variables can be obtained with acceptable precision at the LHU level, as well as more frequent estimates for some regions and the entire country. Inclusion criteria and other methodologic details are reported in the Appendix.
After the sample is extracted, a letter is sent to the homes of the selected individuals. The letter explains the purpose of the system and informs the recipients that they will be contacted shortly; their GPs are also informed. Telephone numbers, if not recorded on the enrollment list, are obtained from telephone directories, from the GPs, or from the respondents themselves when they call to make an appointment after receiving the letter. Survey administrators attempt to call at least 6 times on different days of the week (including weekends) and at different times of the day; if a person cannot be reached, a substitute of the same sex and from the same age group is randomly extracted.
The questionnaire is administered via telephone interviews conducted by specially trained personnel from the public health departments of the LHUs. After briefly explaining the objectives to the respondent, the interviewer obtains oral informed consent. The interviewer uses computer-assisted telephone interviewing or printed questionnaires and subsequent data entry on a personal computer. All data are self-reported. Because they are anonymous, they cannot be validated at the individual level.
The ongoing surveillance process, with interviews conducted every month, provides the system with flexibility, since items can be modified over time; it also permits evaluation of trends and seasonal variations. For small areas, more precise estimates can be obtained by aggregating data over time.
Data collected on the printed questionnaire are entered on a personal computer. All the records are encrypted and transmitted to a centralized database, where basic data-quality controls are performed. No personal identifiers appear in the database. PASSI interviewers and supervisors have protected access to the server through a web portal (www.passidati.it/), with individual user names and passwords and differentiated access profiles. Uploaded records are accessible to the local coordinators, and input errors can be corrected online.
The quality of the data collection process is automatically monitored using indicators modeled after international standards (13,14) (Appendix). After all interviews collected in a calendar year are uploaded, data are checked and edited at the central level.
To calculate regional and national pool estimates, the data from the different LHUs are aggregated. Because the LHUs differ considerably in terms of population size, and the sizes of the samples also differ substantially, a weight specific for each LHU stratum is added to each record to account for the number of interviews performed in each of the 6 strata of the LHU’s sample and the size of the corresponding strata in the LHU’s target population. Edited datasets are subsequently made available for downloading to regions and LHUs. CNESPS performs national-level analyses and provides the regions and LHUs with tools for obtaining local results, such as data analysis plans and codes to run statistical programs.
Data collection began in 2007 in 19 regions, and 20 regions participated in 2008. The number of participating LHUs was 124 (74%) in 2007 and 136 (84%) in 2008. Approximately 85% of Italy’s 18- to 69-year-old population was covered in 2008.
The number of interviews performed rose from 22,006 in 2007 to 37,819 in 2008, of which 13% and 22%, respectively, were computer-assisted telephone interviews. The monthly number of interviews performed in each region varied widely (range, 25-500) because of differences in population size and number of LHUs. The median duration of interviews for both years was 20 minutes.
The demographic composition of the PASSI national pool sample in 2007 reflected the distribution by age and sex of the population in the corresponding LHUs, as determined by the National Statistics Institute. In the PASSI sample, the younger strata were slightly underrepresented and the older strata overrepresented; the maximum difference was 1.1 percentage points compared with official demographic figures (Table).
The eligibility rates for 2007 and 2008 were 92% and 91%, respectively, and the response rate for both years was 82%. The refusal rate was 10% for 2007 and 8% for 2008. Partial interviews and break-offs (Appendix) accounted for less than 1% of all registered interviews.
The National Centre for Disease Prevention and Control (Centro Nazionale per la Prevenzione e il Controllo Delle Malattie, or CCM) provided the National Institute of Public Health with €1 million (US $1.27 million), used for the national coordinating activities and for supporting the regions in the first 3 years of the project. The participating regions and LHUs contributed part-time personnel, and some regions provided incentives for interviewers. The cost of an interview at the LHU level was an estimated €20 (US $25.46), including preparatory activities and data entry but excluding the coordinating costs.
Preliminary results were presented at the 5th International Conference on Behavioral Risk Factor Surveillance in 2007, in Rome, organized by the National Institute of Public Health and the CCM. The national results of the first year of surveillance (2007) were presented in 2008 during an official workshop in Rome. In September 2009, updated PASSI data were presented at a national conference of the Italian program Guadagnare Salute, sponsored by the Ministry of Health. General annual reports and brief reports on specific health topics have been produced and made available to stakeholders. One peer-reviewed article has been published (15). Most regions and many LHUs have already released their surveillance data and used them for local public health activities. National, regional, and local reports, documents, and other relevant material for PASSI are accessible online (www.epicentro.iss.it/passi/english.asp).
Although we are conducting a more complete evaluation of PASSI, following internationally accepted guidelines (16), and validation studies are being performed to compare the results with those provided by other information sources, this article nonetheless highlights some important characteristics of PASSI. First, the study sample appears to be representative of the target population. Second, acceptability by respondents also appears high, as suggested by low refusal rates and the negligible proportion of incomplete interviews. The response rate compares favorably with rates reported in the most successful international surveys (17,18). Furthermore, data collected by PASSI seem to be of practical use, as demonstrated by the fact that many public health agencies use the data for communication, policy making, and health intervention planning.
Perhaps the most important feature of PASSI is that it is carried out by the public health system at the local level, which seems to have many advantages. In particular the sampled individuals were contacted and interviewed by LHU personnel, which seems to have contributed to the high level of participation; LHUs’ lists of residents were readily available for sampling; and direct involvement of LHU personnel in managing the system provided them with the opportunity and motivation to identify and monitor the needs of their populations and their perceptions of the preventive interventions offered.
However, PASSI does have some limitations. In particular, as with any health survey, the reliability and validity of self-reported behavior may be problematic (19). Furthermore, people who are younger than 18 years or older than 69 years are not included in the survey. The age range of 18 to 69 years was chosen because minors cannot provide informed consent to be interviewed by telephone and because even mild cognitive impairment, which is present in a sizable proportion of older people (20), can reduce the reliability of the answers. Moreover, many preventive programs monitored by PASSI, such as tumor screening, are recommended for people up to age 70 years. To obtain information on the health status of the elderly, a different approach (known as PASSI d’Argento) is being piloted with assistance from the national coordinating group; this approach focuses more on health issues of the elderly and allows for face-to-face and, if necessary, proxy interviews.
A critical aspect of PASSI is sustainability. The system is still considered experimental, without a stable institutional setting. Local surveillance is conducted by public health staff who have other routine responsibilities. This situation generates fatigue, which is exacerbated by insufficient political and financial support in some regions, where public health infrastructures are more fragile. Another challenge is maintaining a sense of usefulness when prevalences for many of the variables change little over time. Geographic comparisons have proved useful in sustaining the momentum of the US system (8,9), and as the number of collected interviews increases over time, subgroup analyses at the local level will be possible, improving targeting of health interventions.
Despite these limitations, many regions and LHUs support the goal of creating an institutional setting for PASSI. Furthermore, in 2009, the CCM renewed funding for PASSI for 2 years so that it could be implemented throughout the entire country. Another substantial contribution to the improvement of PASSI has been its link to the international health promotion and surveillance community (21). Our experience shows that a surveillance system such as PASSI can operate with a reasonable amount of resources, producing useful data and gaining the support of its main stakeholders. We believe that PASSI is an interesting model both for countries intending to develop similar surveillance systems and those in which such systems are already in place.
We are grateful to Mark Kanieff for editorial assistance and to all the local and regional PASSI coordinators and interviewers for their competence and commitment.
The CCM of the Italian Ministry of Health provided funding for this study (grant no. 4393/2004-Ccm).
Corresponding Author: Stefano Campostrini, DStat, PhD, Department of Statistics, Ca’ Foscari University, San Giobbe, Cannaregio 873, 30121 Venice, Italy. Telephone: 39-041-2347427. E-mail: firstname.lastname@example.org.
Author Affiliations: Sandro Baldissera, Nancy Binkin, Valentina Minardi, Giada Minelli, Gianluigi Ferrante, Stefania Salmaso, National Institute of Public Health, Rome, Italy; the PASSI Coordinating Group (Paolo D’Argenio, Barbara De Mei, Gabriele Fontana, Alberto Perra, Valentina Possenti, National Institute of Public Health, Rome, Italy; Nicoletta Bertozzi, Department of Public Health, Cesena, Italy; Giuliano Carrozzi, Department of Public Health, Modena, Italy; Angelo D’Argenzio, Department of Public Health 2, Caserta, Italy; Pirous Fateh-Moghadam, Provincial Agency for Health Services, Trento, Italy; Massimo Oddone Trinito, Department of Public Health C, Rome, Italy; Stefania Vasselli, Ministry of Work, Health, and Social Policy, Rome, Italy; Eva Benelli, Stefano Menna, Zadig Scientific Communications, Rome, Italy).
Table. Representativeness of Sample for Italian Behavioral Risk Factor Surveillance System (PASSI), 2007
Appendix. Eligibility and Outcome Rates for the Italian Behavioral Risk Factor Surveillance System (PASSI)
In Progress by Local Health Units Towards a Healthier Italy (Progressi Delle Aziende Sanitarie per la Salute in Italia, or PASSI), the target population consists of all 18- to 69-year-olds who reside in the local health unit (LHU) area. The following equations were used to calculate interview outcomes:Eligibility Rate =
(I + P) + R + (eNC • NC + ½TNF)
(I + P) + R + RNE + (NC + TNF)Response Rate =
(I + P)
(I + P) + R + (eNC • NC + ½TNF)Refusal Rate =
(I + P) + R + (eNC • NC + ½TNF)
The survey (eligible) population consists of residents aged 18 to 69 years enrolled on the LHU lists who have a telephone number available and can be interviewed by telephone.
The ineligible units in the LHU lists may be either people who no longer belong to the target population (eg, who moved away or died; in 2007 they constituted 2% of the sample) or people who are in the target population but are not eligible according to the protocol, such as those who do not understand Italian, who cannot participate in the interview (eg, because of serious handicaps), or who are hospitalized or institutionalized (3% in 2007).
People who are reached by the survey and discovered to be not eligible (RNE) are excluded and replaced by people who are randomly selected from the same age and sex stratum.
Sampled people who are not reached by the survey are substituted following the same procedures; their eligibility status is unknown and must be estimated indirectly. They belong to 1 of 2 categories:
1) Those for whom a telephone number was not found (TNF) despite an exhaustive search following the protocol procedures (8% of the sampled population in 2007). It was assumed that half of these people actually had telephones and were thus eligible, given the proportion of Italian families who do not have a landline or a cell phone (3.9%), as estimated by the National Statistics Institute (www.istat.it/salastampa/comunicati/non_calendario/20070724_01/testointegrale.pdf).
2) Non contacts (NC): sampled people who have a telephone number available but who could not be contacted, despite repeated attempts, following the protocol procedures. The proportion of eligible units in the non contacts (eNC) is estimated from the proportion of eligible units in the reached population.
Contacts are people who are reached by the survey and confirmed to be eligible. Contacts who refuse to be interviewed or who break off the interview (<50% of applicable questions answered) are counted as refusals (R). Contacts who agree to be interviewed are counted as responses. Interviews are considered complete (I) if contacts answer at least 80% of the applicable questions and partial (P) if they answer 50% to 79% of the applicable questions.
The eligibility rate is defined as the number of eligible units divided by the total number of sampled units, eligible and ineligible. The response rate is the number of complete and partial interviews divided by the number of all potentially eligible units in the sample; it resembles the American Association for Public Opinion Research (AAPOR) response rate 4 (13). The refusal rate is the number of refusals and break-offs divided by the number of all potentially eligible units in the sample; it resembles the AAPOR refusal rate (13).
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
This page last reviewed March 30, 2012