This cross-sectional survey was conducted from the early July of 2003 to February, 2004 in Haidian district of Beijing, shortly after the outbreak. Ethics approval for this study was obtained from Peking University Health Science Center Research Ethics Board. The estimation of sample size based on the literature that reported incidence rate of ADRs of the thymosin, interferon or ribavirion [8,9 and 10] according to the following formula: [11], where N is estimation of sample size, α is significant level, tα is the quantile from the standard normal distribution for a two-tailed probability of 1-α, P is the previously reported incidence rate, and d is the absolute deviation from P that could be tolerated. The α and d were predetermined as 0.05 and 0.2, respectively. Given these assumptions, the sample size calculation
revealed that 4,000-5,000 were required to detect the incidence rate of ADRs for four or five medications.

Multi-step, stratified, cluster sampling was used to identify the studied population. First, we divided Haidian district into suburban areas and rural areas. The suburban areas were further stratified on the basis of different occupations, e.g. one stratum for the health care workers (HCWs), one stratum for the government employers. Thus, one or more clusters were randomly selected from each stratum depending on the weights of the stratum that accounted for the whole population. All the people from the randomly selected clusters were included for the survey. For the rural areas, two villages were randomly selected, where all the villagers were selected.

In this study, the investigators consisted of some epidemiologists from public health, pharmacists from SFDA, and general physicians from community hospitals. Before survey, they were all required to attend two-day survey trainings regarding the questionnaire, the techniques of survey, and quality control. Each survey group comprised three investigators, of which one person would be in charge of the quality of the completed questionnaires.

Prior to survey, we developed a brief, self-administered questionnaire, which contained three main sections: socio-demographical characteristics (e.g. age, sex, occupation, occupation), a series of closed and open-ended questions about utilization of medications (e.g. the name of medications, dosage, presence of unusual symptoms), and the translated version of 15-item psychometrics of Impact Event Scale (IES) [12] evaluating the subjects’ concern about SARS. For those subjects who experienced difficulties in completing the questionnaire independently, e.g. the elders, or the people with disabilities, face to face interview were adopted. For those subjects with presence of suspected symptoms related to prophylactic medications use, they were required to fill the form of ADR report, including the date of onset, suspected symptoms, ways of management, and final outcomes. The collected ADRs reports were sent to Beijing Center for ADRs Monitoring and were reviewed by the pharmacists and physicians to determine the causality between the utilization of medications and occurrence of ADRs. The causality judgment included six levels from ‘certain’, ‘probable’, ‘possible’, ‘unlikely’ and ‘conditional/unclassified’ to ‘Unassessable /Unclassified’.

The 15-item self-administered IES, which was developed by Horowitz et al., is a measure of psychological stress after trauma or some events [12]. Surveyed subjects were asked to reported the frequency of symptoms during the epidemic of SARS on a 4-point scale: 0 indicating ‘not at all’, 1 indicating ‘rarely’, 3 indicating ‘sometimes’, and 5 indicating ‘often’ [12]. On the basis of the sum of all scores, suggested cut-off point is 26, above which a moderate or severe impact was indicated [12]. In this study, a person was considered as ‘not concern’ if the total scores ranged from 0-8; ‘little concern’ if scores ranged from 9 to 25, ‘moderate concern’ from 26-43; and ‘serious concern’ if scores were above 44.

The data from the collected questionnaires
were parallel-double entered into the computer using the Epi Info™ 6 [13]. To protect subjects’ confidentiality, their names and other identifiers were deleted. A randomized identification number was assigned for each subject.

All the statistics analyses were performed by using the SPSS software (Version 10.0, SPSS Inc., Chicago, Illinois, USA). Two-tailed test was used and the significant level was predetermined at the 0.05. Logistic regression with backward variable selection method was applied to establish the determinants of prophylactic medications use. Variables examined in the regression model included age, sex, occupation, education, monthly average family income, medical payment plan, history of medications prophylaxis, contact history with SARS cases, and degree of concerns about SARS. The odds ratios (ORs) and adjusted ORs with 95% confidence intervals (CIs) for each studied factor were calculated and presented, respectively.

The subjects with suspected symptoms were considered as ADRs cases only when the causality between the medications use and occurrence of symptoms was judged as ‘certain’, or ‘probable’. The incidence rates of ADRs were reported and compared by age, sex, occupations, and medications use. A trend analysis with χ² statistics was performed to evaluate the trend of incidence rates of ADRs associated with number of medications used.

We distributed approximately 5,000 questionnaires across eleven different stratums, of which 4,401 (88.02%) were completed and collected. The demographic characteristics of the subjects were shown in Table 17.1. Of the 4,401 subjects, 2,571 (58.42) were women, 4,291 (97.5%) with age ≥ 20 years, 2,539(57.69) with high education. Compared with other occupations, there was a relatively higher proportion of HCWs (23.59%) in the surveyed subjects. Of the 4,350 subjects indicated their medical payment plans, 1,627 (36.97%) had free medical services. Of the 4,373 subjects who indicated the history of prophylactic use of medications, 996 (22.63%) subjects admitted that they ever had medications to prevent respiratory diseases in the last three winters. A total of 4,040 (91.80%) reported their concerns about the SARS with completing the translated version of IES (Table 17.1 ).

Of the 4,401 surveyed subjects, 3,426 (77.85%) subjects had prophylactic medications use, of which 1,122 (32.75%) had more than two medications. Of the 3,426 subjects who ever had medications prophylaxis, 2,380 (68.47%) had TCM, 1,089 (31.79%) had thymosin, 329 (9.60%) had flu vaccine, 415 (12.03%) had vitamin, 301 (8.79%) had interferon, 277 (8.09%) had ribavirin, 81 (2.36%) had immunoglobulin, 51 (1.49%) had antibiotics.

The proportions of surveyed subjects with medications prophylaxis by age, sex, education, occupation, monthly income, medical payment plan, history of medications prophylaxis, and degree of concerns about SARS were shown in Table 17.2. The likelihood of prophylactic medications use was significantly associated with sex (being female), certain occupations (e.g. HCWs, administrative
staffs) when compared with homemaker/unemployment, free medical services, positive history of medications prophylaxis, contact history with SARS cases, and degree of concerns about SARS (Table 17.2 ).