A Community Intervention by Firefighters to Increase 911 Calls and Aspirin Use for Chest Pain

Transcription

1 A Community Intervention by Firefighters to Increase 911 Calls and Aspirin Use for Chest Pain Hendrika Meischke, PhD, MPH, Paula Diehr, PhD, Sharon Rowe, BS, Anthony Cagle, PhD, Mickey Eisenberg, MD, PhD Abstract Objectives: To test the effectiveness of an intervention, delivered face-to-face by local firefighters, designed to increase utilization of 911 and self-administration of aspirin for seniors experiencing chest pain. Methods: King County, Washington was divided into 126 geographically distinct areas that were randomized to intervention and control areas. A mailing list identified households of seniors within these areas. More than 20,000 homes in the intervention areas were contacted by local firefighters. Data on all 911 calls for chest pain and self-administration of aspirin were collected from the medical incident report form (MIRF). The unit of analysis was the area. Firefighters delivered a heart attack survival kit (that included an aspirin) and counseled participants on the importance of aspirin and 911 use for chest pain. Main outcome measures were 911 calls for chest pain and aspirin ingestion for a chest pain event, obtained from the MIRFs that are collected by emergency medical services personnel for 2 years after the intervention. Results: There were significantly more calls (16%) among seniors on the mailing list in the intervention than control areas in the first year after the intervention. Among the seniors who were not on the mailing list, there was little difference in the intervention and control areas. The results were somewhat sensitive to the analytical model used and to an outlier in the treatment group. Conclusions: A community-based firefighter intervention can be effective in increasing appropriate response to symptoms of a heart attack among elders. ACADEMIC EMERGENCY MEDICINE 2006; 13: ª 2006 by the Society for Academic Emergency Medicine Keywords: acute myocardial infarction, emergency medical services, aspirin, intervention, education Heart disease is the leading cause of death for American men and women. 1 The two most critical actions that a patient can take when faced with acute myocardial infarction (AMI) symptoms are 1) seeking prompt medical care, preferably by calling 911, so that therapy can begin as soon as possible, and 2) taking an aspirin at the onset of symptoms. 2,3 However, many AMI patients do not follow these life-saving From the Department of Health Services (HM, PD), Department of Biostatistics (PD), and Department of Medicine (ME), University of Washington; and King County Emergency Medical Services Division (SR, AC), Seattle, WA. Received August 10, 2005; revision received October 14, 2005; accepted October 15, Supported by a grant from the National Heart Lung and Blood Institute (R01 HL ). Address for correspondence and reprints: Hendrika Meischke, PhD, MPH, Department of Health Services, Box , University of Washington, Seattle, WA Fax: ; hendrika@u.washington.edu. recommendations. The objective of this study was to test the effectiveness of an intervention, delivered faceto-face by local firefighters, to increase utilization of 911 and self-administration of aspirin for seniors experiencing chest pain. Thrombolytic therapy and emergency angioplasty can alter the course of AMI, reducing morbidity and mortality. 4 6 However, their efficacy decreases with increasing time between symptom onset and treatment. 4,6 8 Patient delay accounts for most of the delay in receiving thrombolytic therapy Quick use of emergency medical services (EMS) for AMI symptoms can dramatically shorten overall out-of-hospital delay time 13,14 and has been associated with early reperfusion therapy. 15 However, more than half of all patients evaluated in emergency departments for possible AMI do not call ,17 Experimental and clinical investigations have shown that early administration of aspirin during an AMI significantly reduces AMI mortality Medical administration of aspirin for AMI has steadily increased over the past decade, as new trials provide increasing evidence of the benefits for AMI patients. Although aspirin ª 2006 by the Society for Academic Emergency Medicine ISSN doi: /j.aem PII ISSN

2 390 Meischke et al. COMMUNITY INTERVENTION BY FIREFIGHTERS for patients with a suspected AMI is now routinely administered by emergency medical personnel and other health care providers, research shows that this is not done universally, or with all eligible patients. 2,21 23 Thus, there are important reasons that patients should be encouraged to take an aspirin when they first experience symptoms of AMI. Small- and large-scale interventions have been conducted to change patients response to symptoms of AMI, to decrease delay time, and to increase utilization of EMS for AMI patients. 17,24 31 These interventions have been moderately successful, and in some cases, very expensive. As such, we decided to focus our efforts on testing an intervention that would be successful in changing patient behavior, rather than focusing on clinical outcomes. On the basis of extensive focus group research with seniors in King County, Washington, an intervention was developed that included local firefighters as educators. The intervention was based on principles of selfregulatory theory 32 and was designed to affect seniors expectations about the signs and symptoms of a heart attack, emphasizing the fact that chest pain during a heart attack is not always severe. Firefighters were trained to discuss individual barriers to the two recommended coping strategies: calling 911 and taking an aspirin. As part of the intervention, firefighters handed the seniors a heart attack survival kit (HASK), which includes a list of warning signs of heart attack, strong recommendation to call 911 and take an aspirin, one 325-mg uncoated adult aspirin, and space to write medications and allergies and important phone numbers. The HASK was designed in the form of a door hanger and had adhesive strips for permanent placement in the home. METHODS Study Design This study was a community-based randomized controlled trial, conducted between July 2000 and December of It was a collaborative effort between the University of Washington, Public Health Seattle King County Health Department, and all the fire districts within King County, Washington, excluding the city of Seattle. The study received human subjects approval from the University of Washington Human Subjects Division. Study Setting and Population Study areas were created in the following manner. There are a total of 32 fire districts in King County, Washington. We created geographical study areas within each fire district so that all areas would have 25 or more expected calls. One hundred twenty-six areas were created within these fire districts that honored natural boundaries and had approximately the same number of calls in the previous year. This was accomplished on the basis of a combination of geocodes. A geocode is a code used by EMS divisions to indicate a one-quarter-square-mile area within a fire district. This code is used mostly for administrative purposes to track and evaluate 911 calls. We defined a study area as a combination of four or more contiguous geocodes, completely contained within a fire district. These 126 areas were assigned at random to treatment and control status within fire district, as follows: within each fire district, we formed strata on the basis of the number of seniors on the list and randomly assigned half the areas in the stratum to treatment and half to control. If there were an odd number of areas in a fire district, the final area was assigned to either treatment or control status with 50% probability. The identification of the target population came from a mailing list (name, address, and age of person) from a local commercial company. We used age as a proxy for risk for AMI. As such, we obtained a mailing list that contained 48,773 names and addresses of households with an individual older than the age of 65 years that could be matched with a list of geocodes. Because our study areas were created on the basis of geocode information, the study status for these individuals could be identified. The mean number of previous chest pain calls and number of seniors per area were nearly identical in the intervention and control groups. The mean age of individuals in the intervention and control areas was identical (78 years of age), as were proportion of males (39% in control areas and 42% in intervention areas) and proportion of individuals reporting taking daily aspirin (23% in control areas and 24% in intervention areas). Study Protocol Passive-consent letters were mailed to the homes of all eligible people in intervention areas (24,582 homes), indicating that a local firefighter would personally deliver a HASK in the near future unless the participant indicated on a self-addressed postcard that such a visit was not welcomed; 11% of eligible individuals returned the postcard. Two weeks after the passive-consent letters were mailed, local firefighters delivered the kits to the homes of the remaining seniors in King County Washington. More than 300 local firefighters were trained face-toface by study staff to discuss the contents of the HASK and to assess and respond to barriers to calling 911 and taking an aspirin for chest pain. Firefighters emphasized the importance of calling 911 first and taking an aspirin as a subsequent action in response to heart attack symptoms. Firefighters stressed the importance of discussing the less typical symptoms of a heart attack (i.e., nausea, jaw pain) as well as the more common symptoms (i.e., chest pain). Each fire department received a list of names and addresses of eligible individuals in their area who needed to be visited. In total, 21,739 seniors needed to be reached by firefighters in King County. Overall, 96% of homes were visited between June of 2001 and December of For each visit, firefighters were asked to fill out a tracking form indicating whether a contact was made (if an eligible participant was not home, the kit was hung on the doorknob with a letter explaining the kit s purpose) and with whom the visit had taken place. To make the task of home visits more manageable for each of the participating fire districts, only one unscheduled attempt at contacting participants was made for each household. Of the 20,984 households that were visited by a firefighter, almost half (49.2%) received the inperson intervention. In 39.2% of the visits, the kit was hung on the doorknob, and in a small number of cases, the participant refused the visit (1.7%), the participant

3 ACAD EMERG MED April 2006, Vol. 13, No Figure 1. Study flow of patients. was ineligible (i.e., younger than 65 and not a spouse; 2.7%), or the outcome was other (7.1%). Figure 1 presents a flow chart of the study activities and participants. Measures The objective of this randomized trial was to test the effectiveness of an intervention, delivered face-to-face by local firefighters, and to increase utilization of 911 and self-administration of aspirin for seniors experiencing chest pain. Data on calling 911 and self-administration of aspirin were collected from the medical incident report forms (MIRF) completed by paramedics. For chest pain calls, especially for individuals older than 65 years of age, advanced life support personnel are dispatched to the scene. A question on aspirin use for the current event was added to the MIRF to assess the study impact for self-administration of aspirin. We obtained the following information from the MIRFs: names of eligible individuals in 99.8% of cases, addresses in 98.2% of cases, geocode in 95.9% cases, dispatch code in 100% cases, and aspirin use in 52.0% of cases. Data collection began in October 2000 and ended in December Data Analysis The final tally had 24,582 seniors in the 63 intervention areas, and 24,191 seniors in the 63 control areas. Time was divided into four periods: the nine months before the intervention, the six months during the intervention, the first year after the intervention (all of 2002), and the second year after the intervention (2003). The unit of analysis was the area, and the dependent variable was the number of calls per area in 2002, in 2003, and in both combined. The number of calls was calculated separately for seniors on and off the list. The numbers of calls in the intervention and control groups were compared by using linear regression, adjusting for the number of calls in the year before the intervention started and for the number of listed seniors in the area. The fire district also was included in the model as a random effect, because the intervention could have been differentially effective in different fire districts. Analyses were performed by using random effects regression in STATA (STATA Corp., College Station, TX), with the number of calls as the dependent variable, treatment group as the fixed effect of interest, fire district as a random effect, and number of calls before and number of listed seniors as covariates. Analyses were performed with the areas as the unit of analysis, first for calls from seniors who were on the seniors list, and second for callers not on the list. A similar analysis was performed by using Poisson regression. For the aspirin data, the same regression model was used to estimate how many calls included aspirin and also the number of additional aspirin calls that were made after adjusting for the increase in all calls that was caused by the intervention (if any). RESULTS In the two study years (2002 and 2003), there were 11,455 records for persons older than age 65 years with chest pain, and 10,302 could be assigned to one of the 126 study areas. Approximately 38% of the 10,302 were matched to a name of a person on the commercial mailing list. Matching rates were similar for participants in intervention and control areas. A total of 3,899 chest pain calls were made, 2,026 in intervention areas and 1,873 in control areas. For participant flow see Figure 1. The analysis was conducted with the area as the unit of analysis. As such, the tables present chest pain calls on the basis of the average number of calls for each area, multiplied by the total number of intervention (63 areas) or control areas (63 areas), rather than the absolute numbers. Because the absolute number of calls for each area was very small, we believe that multiplying the average number of calls by 63 better shows the intervention effect.

4 392 Meischke et al. COMMUNITY INTERVENTION BY FIREFIGHTERS Table 1 Number of Chest Pain Calls for the Intervention and Control Group (Unadjusted) Group Intervention (I) Control (C) I-C I/C Both years 1,183 1, ,040 1,072 ÿ Both years 1,974 1,985 ÿ All 911 calls 3,157 3, Data refer to the average number of chest pain calls across the intervention and control areas, multiplied by 63. = individuals on the mailing list; unlisted = individuals not on the mailing list. Table 1 shows the average number of calls multiplied by the total number of study areas for intervention or control status, and by whether the call came from a person on the list. For example, in the first evaluation year, there were 637 chest pain calls from listed seniors in intervention areas (computed by taking the average number of calls per intervention area, multiplied by 63) and 548 seniors in control areas, a difference of 89 calls that might be attributable to the intervention. The ratio of intervention to control visits, 637/548, is Differences were smaller for the unlisted calls; in fact, in year two, the control areas had more calls. The regression results are shown in Table 2. For example, the first line shows that in a regression adjusting for the number of calls in the year before the intervention and for the number of seniors in the area, treating the fire district as a random effect, the intervention group had 93 more calls than the control group in This difference was statistically significant (p = 0.04, 2-tailed test; 95% CI = 4 to 182). On the basis of the Poisson regression, the intervention group had 1.16 times as many calls as did the control group, which also was statistically significant (95% CI = 1.03 to 1.30). Among the unlisted seniors who could not have received the intervention, there was very little difference in the intervention and control areas, and the control numbers were actually higher in some comparisons. For the two evaluation years combined, there was a significant difference for the listed seniors and no significant difference for the unlisted seniors. The treatment differences for the calls (both years, listed and unlisted) showed a positive but nonsignificant effect. Table 3 shows the number of calls of each type by treatment group, taken from Table 1. The bottom of Table 3 shows the number of calls of each type where the person had taken aspirin. For example, in 2002, there were 216 listed calls in which the person took aspirin in the intervention group, compared with 163 in the control group, for a difference of 53 calls and a ratio of The percentage of calls with aspirin was 216/ = 33.9% of calls in the intervention group (2002) versus 29.7% of all calls in the control group, a difference of 4.2 percentage points and a ratio of 1.14 (the percentage of calls with aspirin was 1.14 times as high in the intervention group). There thus appears to be a positive effect of the intervention on aspirin taking in 2002, over and above the higher number of people calling 911. In 2003, however, there were 11 fewer aspirin takers in the intervention group than the control group. For persons not on the list, the control group had slightly more aspirin takers in 2002 and This suggested a significant but short-lived effect on persons who were on the list, but no effect on those who could not have received the intervention. The regression results for aspirin use are in Table 4, which shows a significant increase in the number of calls that mention aspirin in 2002 (two-tailed test) and in 2002 and 2003 combined for persons on the list (one-tailed test), but not for 2003 for persons not on the list. The adjusted differences are similar to those in Table 3. Adding the number of calls as a covariate in the regression, the treatment effect was not significant (not shown), suggesting that the increase in reports of aspirin use may have been caused by the increase in the number of calls, rather than a specific aspirin effect. DISCUSSION The results of the HASK trial show that firefighters can be effective educators in the community. Additional analyses (not shown) showed that in the two-year study period after the intervention, 3.7% of the listed persons with face-to-face contact had one or more calls to 911 as compared with only 2.7% of those whose materials Table 2 Regression Results Comparing the Number of Calls in the Intervention and Control Groups Group Adjusted Number of Additional Calls Because of Intervention p-value 95% CI Adjusted Ratio of Intervention to Control p-value 95% CI , , ÿ18, , 1.26 Both years , , ÿ105, , ÿ ÿ210, , 0.99 Both years ÿ287, , 0.98 All calls ÿ201, , 1.06 Data were computed by multiplying the regression coefficient for the intervention group by 63 and were adjusted for the number of calls in the year before the intervention started and for the number of listed seniors in the area.

5 ACAD EMERG MED April 2006, Vol. 13, No Table 3 Number of Calls and Number with Aspirin per Period Group Intervention (I) Control (C) I-C I/C Number of calls (from Table 1) ,040 1,072 ÿ Number of calls with Aspirin ÿ ÿ ÿ Percentage of calls with Aspirin* ÿ ÿ ÿ * (Number of calls with aspirin/number of calls) 100. were left on the doorknob (p = 0.000). In addition, the intervention increased visibility of EMS personnel in many communities, and our research office received many anecdotes of positive experiences for both the seniors as well as the firefighter personnel. LIMITATIONS Although the positive experience of this trial is encouraging, there are some cautionary notes. First, the intervention did not reach all seniors in the intervention areas. The mailing list used for identification of seniors was only about 50% complete. In addition to the incomplete mailing list, the firefighters reached only half of all participants on their list. This is not surprising given the fact that these visits were unscheduled one-time attempts at contact. Because of the public health nature of this large-scale intervention, we decided against scheduling and multiple attempts at visiting. However, because of the labor-intensive nature of this project, it is important to investigate ways in which firefighters can reach a broader section of the senior community (maybe by canvassing senior high-rises, senior centers, or other retirement places). Second, the compliance with data collection on aspirin use was less than optimal. In only 52% of the MIRFs did we have completed records on aspirin use. The question on aspirin use was added to the MIRF for this particular study, and many of the EMS personnel were not used to filling out the question. Because this is a randomized trial, we feel that the low completion rate does not compromise the validity of the results regarding aspirin. Third, the analyses were somewhat dependent on the effect of an outlier community as well as the statistical model used in the analysis. There was one outlier area in the intervention group whose number of calls in 2002 was substantially higher than in the baseline period, as compared with the other areas. If that area was removed from the analysis, the intervention group advantage decreased to 77 calls, and was significant when a one-tailed (but not a two-tailed) test was used. We found out that in this study area, several senior housing complexes were built during the study period, and changes in their 911 policy might have caused some effect. However, this effect did not occur for the unlisted calls, suggesting that the outlier area may not be a problem. Another consideration is that the independence assumptions of the Poisson regression may not have been satisfied, because a person who makes one call is likely to make additional calls. A negative binomial regression showed a treatment effect, but it was not statistically significant. The findings, though statistically significant and potentially large, need verification in other settings, in part because the findings were somewhat sensitive to the analytic method chosen. The HASK program comes with its associated costs, the largest being the $3.50 paid per house call as an incentive and compensation for gas and other expenses. We did NOT pay for salaries or overtime as this was contributed by the fire departments. Nevertheless, even a modest fee of several dollars per visit may be prohibitively expensive to communities, especially in an era of constricting resources. The most realistic way for HASK to be generalizable is for EMS agencies to expand their mission and raison d être to include prevention and education of their service population about lifethreatening disease. Table 4 Regression Results for Aspirin Use in Evaluation Period Group Adjusted Number of Additional Calls Because of Intervention p-value 95% CI Adjusted Ratio of Intervention to Control p-value 95% CI , , ÿ ÿ39, , 1.18 Both years ÿ5, , ÿ51, , ÿ ÿ35, , 1.20 Both years ÿ65, , 1.14 All calls ÿ45, , 1.19

6 394 Meischke et al. COMMUNITY INTERVENTION BY FIREFIGHTERS CONCLUSIONS We believe that campaigns that use an interactive, faceto-face counseling session by authority figures, such as in the HASK study, offer much promise but require a lot of community cooperation. Fire departments and EMS agencies are the logical organizations to take on the responsibility of acute coronary syndrome public education. Although the participation of fire districts in King County, Washington may be unique, our office has received many inquiries from EMS departments nationwide about our study, suggesting that the intervention may be of interest to other EMS departments as well. The authors are grateful for the support this project received from the fire chiefs and firefighter personnel of the participating fire districts in King County, Washington. References 1. American Heart Association. Heart Disease and Stroke. Statistics. Update Dallas, TX: American Heart Association, Hennekens CH, Dyken ML, Fuster V. Aspirin as a therapeutic agent in cardiovascular disease. A statement for healthcare professionals from the American Heart Association. Circulation. 1997; 96: National Institutes of Health. Educational Strategies to Prevent Prehospital Delay in Patients at High Risk for Acute Myocardial Infarction. Bethesda, MD NHLBI No Lundergan CF, Reiner JS, Ross AM. How long is too long? Association of time delay to successful reperfusion and ventricular function outcome in acute myocardial infarction: the case for thrombolytic therapy before planned angioplasty for acute myocardial infarction. Am Heart J. 2002; 144: Marder VJ, Sherry S. Thrombolytic therapy: current status (second of two parts). N Engl J Med. 1988; 318: Goff DC, Nichaman MZ, Ramsey DJ, Meyer PS, Labarthe DR. A population-based assessment of the use and effectiveness of thrombolytic therapy: the Corpus Christi Heart Project. Ann Epidemiol. 1995; 5: Steg PG, Bonnefoy E, Chabaud S, et al. Impact of time to treatment on mortality after prehospital fibrinolysis or primary angioplasty: data from the CAPTIM randomized clinical trial. Circulation. 2003; 108: Meils CM, Kaleta KA, Mueller CL. Treatment of the patient with acute myocardial infarction: reducing delay times. J Nurs Care Qual. 2002; 17: Dracup K, Moser DK, Eisenberg MS, Meischke H, Alonzo A, Braslow A. Causes of delay in seeking treatment for heart attack symptoms. Soc Sci Med. 1995; 40: Goldberg RJ, Steg PG, Sadiq I, et al. Extent of, and factors associated with, delay to hospital presentation in patients with acute coronary disease (the GRACE registry). Am J Cardiol. 2002; 89: Goldberg RJ, Yarzebski J, Lessard D, Gore JM. Decade-long trends and factors associated with time to hospital presentation in patients with acute myocardial infarction: the Worcester Heart Attack study. Arch Intern Med. 2000; 160: Goff DC Jr, Feldman HA, McGovern PG, et al. Prehospital delay in patients hospitalized with heart attack symptoms in the United States: the REACT trial. Rapid Early Action for Coronary Treatment (REACT) Study Group. Am Heart J. 1999; 138: Grim PS, Feldman T, Childers RW. Evaluation of patients for the need of thrombolytic therapy in the prehospital setting. Ann Emerg Med. 1989; 18: Weaver WD, Martin JS, Litwin P, et al. Prehospital thrombolytic therapy MITI Project report on phase I: feasibility, characteristics of patients. J Am Coll Cardiol. 1989; 13:152A. 15. Canto JG, Zalenski RJ, Ornato JP, et al. Use of emergency medical services in acute myocardial infarction and subsequent quality of care: observations from the National Registry of Myocardial Infarction 2. Circulation. 2002; 106: Ho MT. Delays in treatment of acute myocardial infarction: an overview. Heart Lung. 1991; 20: Luepker RV, Raczynski JM, Osganian S, et al. Effect of a community intervention on patient delay and emergency medical service use in acute coronary heart disease: the Rapid Early Action for Coronary Treatment (REACT) Trial. JAMA. 2000; 284: Collins R, Peto R, Baigent C, Sleight P. Aspirin, heparin, and fibrinolytic therapy in suspected acute myocardial infarction. N Engl J Med. 1997; 20: Fuster V, Dyken ML, Vokonas PS, Hennekens C. Aspirin as a therapeutic agent in cardiovascular disease. Circulation. 1993; 87: Hennekens CH, Jonas MA, Buring JE. The benefits of aspirin in acute myocardial infarction. Still a wellkept secret in the United States. Arch Intern Med. 1994; 154: Eisenberg MJ, Topol EJ. Prehospital administration of aspirin in patients with unstable angina or acute myocardial infarction. Arch Intern Med. 1996; 156: McLaughlin TJ, Soumerai SB, Willison DJ, et al. The effect of comorbidity on use of thrombolysis or aspirin in patients with acute myocardial infarction eligible for treatment. J Gen Intern Med. 1997; 12: Barbash IM, Freimark D, Gottlieb S, et al. Outcome of myocardial infarction in patients treated with aspirin is enhanced by pre-hospital administration. Cardiology. 2002; 98: Blohm M, Hartford M, Karlson BW, Karlsson T, Herlitz J. A media campaign aiming at reducing delay times and increasing the use of ambulance in AMI. Am J Emerg Med. 1994; 12: Blohm MB, Hartford M, Karlson BW, Leupker RV, Herlitz J. An evaluation of the results of media and educational campaigns designed to shorten the time taken by patients with acute myocardial infarction to decide to go to hospital. Heart. 1996; 76: Eppler E, Eisenberg MS, Schaeffer S, Meischke H, Larsen MP and emergency department use for chest pain: results of a media campaign. Ann Emerg Med. 1994; 24:202 8.

7 ACAD EMERG MED April 2006, Vol. 13, No Gaspoz JM, Unger PF, Urban P, et al. Impact of a public campaign on pre-hospital delay in patients reporting chest pain. Heart. 1996; 67: Herlitz J, Blohm M, Hartford M, et al. Follow-up of a 1-year media campaign on delay times and ambulance use in suspected acute myocardial infarction. Eur Heart J. 1992; 13: Ho MT, Eisenberg MS, Litwin PE, Schaeffer SM, Damon SK. Delay between onset of chest pain and seeking medical care: the effect of public education. Ann Emerg Med. 1989; 18: Meischke H, Eisenberg MS, Schaeffer S, Henwood D. The Heart Attack Survival Kit project: an intervention designed to increase seniors intentions to respond appropriately to symptoms of acute myocardial infarction. Health Educ Res. 2000; 15: Meischke H, Durlberg EM, Schaeffer SS, Henwood DK, Larsen MP, Eisenberg MS. Call Fast, Call 911 : a direct mail campaign to reduce patient delay in acute myocardial infarction. Am J Public Health. 1997; 87: Leventhal H, Nerenz DR, Steele DJ. Illness representations and coping with health threats. In: Baum A, Singer J, eds. A Handbook of Psychology and Health. Hillsdale, NJ: Erlbaum, 1984, pp