BASTA, TANIA B.1; REECE, MICHAEL2; WILSON, MARK G.3
At the end of 2003, there were over 1 million individuals living with HIV/AIDS in the United States; furthermore, 40,000 Americans become newly infected each year (13). However, although the incidence continues to increase, HIV-related deaths have decreased by 83% in the United States due to the introduction of highly active antiretroviral therapy (HAART) in 1996 (22). To that end, HIV is being considered a chronic disease, such as hypertension and diabetes. Because individuals now have the potential to live upward of 20 yr on HAART, treatment has expanded to include behaviors aimed at maximizing the quality of life among these individuals. Exercise is a self-care behavior, which has been shown to produce benefits in individuals living with HIV that are similar to the general population (24,25,36,38,45,48). Furthermore, it is the most often cited form of complementary and alternative therapy among individuals living with HIV (16), yet currently, there are few published community-based exercise behavior or intervention studies.
Although individuals living with HIV/AIDS have the potential to live longer and healthier on HAART, there are side effects that affect quality of life, both psychologically and physically. Some of the common side effects include fatigue, anemia, digestive problems, diarrhea, skin rashes, neuropathy, bone problems, and body fat redistribution (31,43). Furthermore, psychological symptoms such as anxiety, depression, and emotional distress have been associated with early and late stages of HIV disease (6,35,47). These side effects, coupled with the diagnosis of a potentially fatal disease, may affect one's intention and motivation to perform regular exercise. In an effort to develop interventions for individuals living with HIV/AIDS, it is important for researchers to examine exercise behavior in this population using current health behavior theories, including the transtheoretical model (TTM) of behavior change.
The TTM was developed to assess how people intentionally change behavior (37). This theory suggests that individuals modify their behavior by moving sequentially through five stages of change: precontemplation (not intending to make a change), contemplation (considering making a change), preparation (making small changes), action (actively engaging in behavior change), and maintenance (sustaining change over time). Movement through the stages of change is influenced by three constructs: 1) decisional balance, 2) processes of change, and 3) self-efficacy.
The TTM has been applied to numerous health behaviors because it allows researchers and practitioners the ability to tailor interventions to match an individual's stage of change. Furthermore, the TTM allows for the measurement of change from one stage to another. During short interventions, individuals may not report a change in behavior but may report movement from one stage to the next, indicating a change in motivation. To that end, even if an intervention does not create a documentable behavior change, the stage movement suggests that a change in behavior might occur in the future.
The TTM has been applied to exercise behavior predominately in worksite (7,19,26-28,30), university (7,29), and community settings (8,9,17,34,39,41,46,50). Results from these studies indicate that three of the TTM constructs, self-efficacy, the processes of change, and decisional balance, are effective at moving individuals through the stages and that they can be used to predict the stage of change for exercise (9) among individuals in these diverse settings. Currently, there is only one published study that has assessed the TTM for exercise behavior in individuals living with HIV/AIDS (4), and it was conducted as part of the current study.
Although the TTM is good at detecting change in motivation to alter a behavior, recent review articles have suggested that there is limited evidence to support the use of the TTM for individual behavior change (1,2,5,44,49). Some have cited methodological issues, such as threats to internal and external validity, and conceptual problems, such as limitations of the utility of the TTM. One of the most frequently cited issues with the TTM has been related to the validity of the staging instruments. According to the model, staging individuals is crucial for program delivery, but few validated algorithms exist creating a threat to internal validity (1,2). As a result, if a person is not staged appropriately, then the targeted intervention may not address their motivational readiness to change.
Predictive discriminant analysis (PDA) is a statistical analysis that is used to estimate the predictive power of a set of variables (21). The independent variables in the analysis are used to develop a rule to predict group membership or to classify individuals into groups. Therefore, the PDA approach can be used to determine 1) the most important TTM predictors of stage of change, 2) how often the TTM constructs predict an individual's stage of change, 3) whether the prediction is statistically better than chance alone, and 4) a prediction rule for future participants. Discriminant analyses have been used in five recent studies involving the TTM for exercise behavior (9-11,18,23). One study included only the composite scores for self-efficacy, decisional balance, experiential processes of change, and behavioral processes of change as predictors (23), whereas the other studies also included barriers to exercise (9,18) and demographic variables (10,11) as predictor variables.
In the five aforementioned studies, "predictive discriminant analysis" was not cited as the primary statistical analysis; however, an overall hit rate was reported, which suggests that a mixture of descriptive and predictive discriminant analyses were used, although not recommended (20). The overall hit rate of the previous studies or the number of individuals classified into the "correct" stages (21) ranged from 47.9% to 69.6% (9,10,18,23). Two studies of Cardinal et al. (9,10) and one Kosma et al. (23) reported the individual hit rates by stage. In the study of Kosma et al. (23), only the TTM constructs were used as predictors. The stages that were most accurately predicted were contemplation (76.3%), preparation (58.3%), and precontemplation (40%), whereas the least accurately predicted stages were maintenance (8.3%) and action (0%). In one of the studies of Cardinal et al. (9), the sample consisted of individuals living with physical disabilities. In that study, the predictor variables included the TTM constructs as well as barriers to exercise. The stages that were predicted most accurately were maintenance (91.3%) and precontemplation (73.8%), and the least accurately predicted stages were contemplation (48.3%), preparation (23.8%), and action (5.3%). In the other study of Cardinal et al. (10), the sample consisted of Finnish and American college students, and the predictor variables included the TTM constructs as well as five additional variables: nationality, gender, age, body mass index (BMI), and exercise METs. The stages that were predicted most often were maintenance (88.8%) and contemplation (71.7%) compared with precontemplation (30.8%), preparation (27.8%), and action (26.2%), that were the least accurately predicted stages. Most importantly, none of these studies reported whether the prediction of the stages was better than chance alone.
The current study is part of a larger study, that was the first published study to apply the TTM for exercise among individuals living with HIV/AIDS (4). The purpose of the larger study was to examine 1) the distribution of the stages of change, 2) the differences in the TTM constructs explained by the stages of change, and 3) the relationship between physical activity and the stages of change among individuals living with HIV. The entire study was developed in collaboration with practitioners at a community-based HIV-related organization who were interested in developing a stage-matched exercise intervention for their clients. This study was developed to determine whether the TTM instruments were valid in this special sample as well as whether the clients were staged accurately. If the clients were not staged better than chance alone, then there would be little support for using a stage-matched approach over a nontailored approach. Therefore, the purpose of this study was to assess 1) the overall hit rate or the number of individuals classified into the correct stages of change for exercise behavior, 2) the individual hit rates for each of the five stages of change for exercise behavior, and 3) whether the TTM constructs (decisional balance, self-efficacy, experiential processes of change, and behavioral processes of change) for exercise behavior predicted the stages of change for exercise significantly better than chance alone among individuals living with HIV.
This study used a cross-sectional design in which data were collected from four HIV-related community-based organizations in a large southeastern city. This study was approved by the Institutional Review Board (IRB) at the University of Georgia.
An a priori multivariate-level power analysis (42) determined that 170 participants were needed to ensure the internal validity of the results (i.e., effect size [d] ≥ 0.50, α ≤ 0.05, power = 0.80). A total of 208 individuals living with HIV agreed to participate in this study. Most of the sample self-identified as male (87.0%), African American or black (84.6%), single (69.2%), and homosexual (54.8%). More than half (54.3%) of the sample was unemployed due to disability, and 85.1% of the sample held at least a high school diploma. The ages ranged from 22 to 63 yr with an average age of 42.6 yr (SD = 7.1).
The participants had been living with an HIV diagnosis for an average of 11.75 yr (SD = 6.59) but ranged from 2months to 25 yr before the study began. For those individuals who had also received an AIDS diagnosis (41.3%), the average time since AIDS diagnosis was 6.59yr (SD = 4.86) but ranged from 1 month to 20 yr before the study began. The CD4 count, a measure of immune function, ranged from 3 to 1697 cells·mm−3 with an average of 433.77 cells·mm−3 (SD = 267.00) for the 154 participants who responded to the CD4 question. A healthy person's CD4 count ranges from 800 to 1200 cells·mm−3, and an AIDS diagnosis occurs when a person's CD4 count drops below 200 cells·mm−3. There were no significant differences in stages of change among individuals who had received an HIV diagnosis and those who had also received and AIDS diagnosis.
To measure decisional balance for exercise, a 10-item scale was used (32). Five items measured the pros for exercise, and five items measured the cons for exercise. All items were assessed on a five-point Likert scale (1 = not at all important, 2 = a bit important, 3 = somewhat important, 4 = quite important, and 5 = extremely important). The current study yielded an internal consistency (15) of 0.82 for the pro scale and 0.76 for the con scale.
The processes of change were measured by a 30-item scale developed by Nigg et al. (33). Fifteen items assessed the behavioral processes of change (i.e., contingency management, counterconditioning, helping relationships, self-liberation, and stimulus control), and 15 items assessed the experiential processes of change (i.e., consciousness raising, dramatic relief, environmental reevaluation, self-reevaluation, and social liberation). All items were measured on a five-point Likert scale (1 = never, 2 = seldom, 3= occasionally, 4 = often, and 5 = repeatedly). In the current study, the subscales for the experiential processes of change ranged from 0.59 to 0.78, and the subscales for the behavioral processes of change ranged from 0.73 to 0.85.
Self-efficacy for exercise was measured using a six-item measure developed by Marcus et al. (27). This scale assessed six components that predict exercise behavior: negative affect, excuse making, exercising alone, inconvenient to exercise, resistance from others, and bad weather. Each item was measured on a five-point Likert scale (1 = not at all important, 2 = a bit important, 3 = somewhat important, 4 = quite important, and 5 = extremely important). In the current study, the internal consistency of these measures was 0.76.
An individual was classified into one of the five stages of change by their response to a one-item staging algorithm developed by Marcus et al. (27). For this item, regular exercise was defined as, "Any planned physical activity (e.g., walking, jogging, bicycling, swimming, dancing, tennis, rowing, weight lifting, etc.) performed to increase physical fitness. Vigorous activity is hard physical effort that makes you breathe much harder than normal and should be performed 3 or more times per week for 20 or more minutes per session. Moderate activity is moderate physical effort that makes you breathe somewhat harder than normal and should be performed 30 or more minutes a day five days a week," which is consistent with the ACSM (3) and CDC (14) exercise guidelines. This algorithm has demonstrated concurrent validity with the 7-d Physical Activity Recall Questionnaire (40). A response of "Yes, I have been exercising for more than 6 months" categorized individuals in the maintenance stage. A response of "Yes, I have been exercising for less than 6 months" classified an individual in the action stage. A response "No, but I intend to in the next 30 d" categorized individuals in the preparation stage. A response of "No, but I intend to in the next 6 months" classified individuals in the contemplation stage. A response of "No, and I do not intend to in the next 6 months" categorized individuals in the precontemplation stage.
Data were collected from February to April 2006 in an urban southeastern city. Recruitment flyers were posted in the common areas of the four HIV-related service organizations. Recruitment materials clearly indicated that participants were not required to be active exercisers. All participants had already received an HIV diagnosis and were currently enrolled in HIV care services at community-based organizations funded by the Ryan White Care Act. All participants gave verbal consent to participate, then completed the questionnaire anonymously, and were compensated $5 in cash for their participation.
On the basis of responses to the staging algorithm, participants were classified into one of the five stages of change for exercise: 1) precontemplation, 2) contemplation, 3) preparation, 4) action, or 5) maintenance. Descriptive analyses were conducted using version 14.0 of the Statistical Package for the Social Sciences (SPSS, Inc., Chicago, IL).
A direct predictive discriminant analysis (PDA) was used to estimate the predictive power of a set of variables. The predictor variables included scores for 1) self-efficacy, 2) decisional balance, 3) the behavioral processes of change, and 4) the experiential processes of change. The outcome variables were the five stages of change: 1) precontemplation, 2) contemplation, 3) preparation, 4) action, and 5) maintenance.
There are two assumptions in PDA: 1) multivariate normality and 2) equal variance-covariance matrices (21). The normality of each variable distribution was checked by examining the corresponding box plot. Results from Box's test indicated that covariance matrices were unequal; therefore, a quadratic classification was applied and SAS was used for the data analysis. The probability that individuals were classified into each stage was based on prior probabilities on the basis of a national estimate: 14% in precontemplation, 14% in contemplation, 29% in preparation, 7% in action, and 36% in maintenance (12).
The hit rate, or the number of individuals classified into the "correct" stages, was calculated using an external cross-validation classification analysis, the "quadratic leave-one-out" method (21). This method developed a prediction model on the basis of N−1 and then tested the classification prediction equation on the one person left out of the analysis. This analysis was repeated for each person in the sample, and then the hit rate was calculated on the basis of the number of people correctly predicted into their known stages.
The overall expected hit rate (chance) was calculated, and then the observed hit rate was compared with the expected hit rate to see whether there was a statistically significant difference between chance alone and the prediction equation. Finally, "improvement over chance" was calculated to determine how much better the prediction of stages was over chance alone. In addition to the overall hit rates, an expected and observed hit rate were compared for each stage, and the "improvement over chance" calculation was conducted for each stage as well.
Stages of Change
All 208 participants self-selected into one of the five stages of change. On the basis of the participants' responses to the one-item staging algorithm, 23 (10.8%) were classified in the precontemplation stage, 32 (16.8%) in the contemplation stage, 33 (15.6%) in the preparation stage, 49 (23.6%) in the action stage, and 71 (33.5%) in the maintenance stage.
Predictive Discriminant Analysis
The 5 × 5 classification table is presented in Table 1. This table presents the number of individuals classified into each of the five stages. The five separate-group hit rates or number of individuals classified in the correct stages are given in parentheses on the main diagonal. The total group hit rate was 42%, which was statistically better than chance alone (Z= 6.79, P < 0.05) or 25% better than what may be expected by chance. The precontemplation group hit rate was 48%, which was statistically better than chance alone (Z = 4.69, P < 0.05) or 40% better than what may be expected by chance. The hit rate for the contemplation group was 25%, which means the hit rates were no better (Z = 1.80, P > 0.05) than what may be expected by chance. The hit rate for the preparation group was 70%, which was statistically better than chance alone (Z = 5.15, P < 0.05) or 58% better than what may be expected by chance alone. The hit rate for the action group was 0, which means that no individuals in action were predicted by the TTM constructs. Finally, the maintenance group hit rate was 63%, which was statistically significant (Z = 4.66, P < 0.05) or 42% better than what may be expected by chance alone. Table 2 presents a comparison of the hit rates by stage of change from this study and three other studies (9,10,23).
Classification rules for new scores.
The quadratic classification functions (QCF) for precontemplation are presented in Table 3. Due to limited space, the QCF for all five stages are not included. On the basis of the QCF, classification equations were developed that would help predict an individual's stage on the basis of their overall decisional balance, self-efficacy, experiential processes of change, and behavioral processes of change composite scores. To determine which stage an individual should be classified in, the scores for self-efficacy (X1), decisional balance (X2), experiential processes of change (X3), and behavioral processes of change (X4) were entered into all five equations, and the equation with the highest score became the stage to which the individual should be classified.
The predictive discriminant analysis classified individuals into the correct stages 42% of the time, which is statistically better than chance alone. However, it also indicates that 58% of the individuals in this study were classified incorrectly. Other studies using PDA for exercise correctly classified between 47.9% and 69.4% of individuals (9-11,18,23), so individuals in this study were accurately classified less often than that in five other studies. Results of this study suggest that participants either self-identified into the wrong stage or they did not answer the decisional balance, self-efficacy, or processes of change items in a consistent manner with the stage they chose. This is an important finding because the TTM posits that on the basis of an individual's self-reported stage, a tailored intervention using the processes of change, decisional balance, and self-efficacy can be matched to their readiness to change. An intervention can incorporate activities or messages that target how an individual is thinking and behaving on the basis of their particular stage of change. Therefore, if an individual is not staged appropriately, then intervention messages will not match their motivation to change and, as the model suggests, will not be as effective in changing behavior.
One possible explanation for the lower overall hit rate is that in this study the authors only included the TTM constructs as the predictor variables. Kosma et al. (23) included the TTM constructs as predictors, and the overall hit rate was 54.3%, which is higher than that in this study, but lower than that of other studies, which included other variables as predictors (e.g., BMI, METs, social support, barriers to exercise). However, the authors of this study were interested in assessing the internal validity of the TTM and therefore did not want to include other variables as predictors.
The precontemplation, the preparation, and the maintenance stages all were predicted at rates significantly better than chance alone. The preparation stage was the most accurately predicted stage with 70% of participants being accurately classified in this stage. These findings contradict previous studies in which they found that preparation was only accurately predicted 23.8% (9) and 27.8% (10) of the time. However, Kosma et al. (23), who included individuals living with physical disabilities, accurately staged 58.3% of their participants in the preparation stage, which is closer to the 70% accurately staged in this study than the other two studies. One possible explanation is that individuals in this study and in the study of Kosma et al. (23), were living with a chronic disease or physical impairment and were seeking out treatment for their condition. Therefore, going for treatment might have served as a "gateway" for learning about self-care behaviors, including exercise. As a result, individuals in this study, as well as in the study of Kosma et al. (23), may have recently learned of the benefits of exercise and may have just decided they would start exercising in the next 30 d.
The precontemplation and the maintenance stages were the most accurately predicted stages in the study of Cardinal et al. (9) and were the most reliably predicted in this study behind the preparation stage. The precontemplation stage in this study was predicted 48% of the time, whereas it was predicted 73.8% of the time in the study that examined individuals living with physical disabilities (9). Furthermore, the maintenance stage in this study was predicted 68% of the time in this study and was predicted91.3% and88.8% of the time in the studies of Cardinal et al. (9,10). A possible explanation for precontemplation and maintenance being predicted more reliably than other stages is that individuals in these stages either have been exercising for more than 6 months or are not thinking about beginning exercise in the next 6 months. They are not in the contemplation or preparation stages where they might be thinking about beginning to exercise regularly. Therefore, their attitudes and beliefs about exercise probably align most consistently with their self-reported stage of exercise.
The predictive discriminant analysis did not classify anyone correctly into the action stage in this study, which is consistent with the study of Kosma et al. (23) in which no participants were classified correctly into the action stage. Furthermore, only 5% of participants living with physical disabilities were accurately classified into the action stage (9), and the action stage was also the least accurately predicted stage among the Finnish and American college students with only 26.2% being accurately predicted (10). This finding may be because exercise, although it can be an addictive behavior, is not as addictive as nicotine and other drugs. The TTM was initially developed to explain the processes smokers undertake as they quit smoking (37). When an individual enters the action stage for smoking cessation, there is an actual definitive day when the smoking behavior ends. However, there is not always a clear demarcation of when the regular exercise begins or what one considers regular exercise. Furthermore, the length of the action stage may not be appropriate for exercise. Individuals who just began performing regular "planned" exercise may still think like a person in preparation, whereas a person who has been in the action stage for 5 months may think more like a person in maintenance. Or, perhaps, as previous studies have reported (9,10,23), it is often difficult to cognitively distinguish individuals in action from maintenance, which might explain why individuals in this study were classified most often in maintenance and not in action. Another possible explanation is that the instruments used in this study were validated in healthy populations, and the predictive factors identified may not generalize to the current sample.
The participants in this study were unique. They were all low-income and living with HIV/AIDS as well as predominately African American, male, homosexual, unemployed, and high school graduates. Given the vast variability among individuals living with HIV, the results of this study may be generalizable to individuals with similar characteristics but not generalizable to all individuals living with HIV/AIDS. Furthermore, the study used a cross-sectional design; therefore, no causal or temporal statements can be made about the relationship between the stages of change for exercise and the other TTM constructs (decisional balance, self-efficacy, and the behavioral and experiential processes of change). Furthermore, there may be something special about the particular point in time when individuals in this sample completed the questionnaire. Individuals went to the study sites for HIV-related services and perhaps may have been sick or seeking out treatment education or information on exercise and nutritional supplements. To that end, a cross-sectional design may have given a distorted picture of the typical exercise behavior that this population engages in on a regular basis.
Three of the experiential processes of change had reliability scores that were between 0.59 and 0.63, which is less than desirable. If these processes have low reliability in future studies, then the wording on the individual items may need to be altered for this unique population. For example, the counterconditioning process has an item that asks, "Instead of taking a nap after work, I exercise." Because most of this sample was unemployed, this item may not have been an appropriate item to measure counterconditioning in this study.
Implications for practice.
For three of five stages, participants were classified into stages better than if done by random chance. Therefore, the results suggest that using the TTM for audience segmentation among individuals living with HIV may be a better approach than nonsegmented mass communication approaches. As a result, it appears that the TTM should be used to develop targeted messages about exercise. However, the results also highlight the importance of making sure that the staging algorithm is accurately staging individuals living in an appropriate stage on the basis of the other TTM constructs.
Implications for research.
More studies need to conduct predictive discriminant analyses to see whether individuals are classified into the proper stages. One of the criticisms of the TTM is the inability to accurately stage participants, yet PDA is a statistical analysis that can predict the accuracy of the stage classification based on composite scores for self-efficacy, decisional balance, and the processes of change. Furthermore, once prediction equations are developed, they should be validated using new participants to ensure that the prediction equations were not overfitted to the development sample.
The results also suggest that perhaps action is not a feasible stage for exercise among individuals with disabilities. Among four of the studies, the TTM constructs either did not predict individuals into the action stage or their prediction was the poorest for this stage. This is an interesting commonality among individuals living with disabilities that should be explored further in future studies.
Future studies also need to include other variables, in addition to the TTM constructs that theoretically might affect stage prediction to see whether the accuracy of the stage prediction is improved. The studies of Cardinal et al. (9,10) included barriers, BMI, METs, and demographic variables in addition to the TTM constructs but did not report the hit rate with only the TTM constructs as predictors. However, for the study that was conducted among individuals living with physical disabilities, the overall hit rate only decreased by 1.2% when the barriers variable was taken out of the analyses (9). Perhaps including other variables such as barriers and social support for exercise might increase the predictive ability of the constructs.
In conclusion, this is only the second known published study that applied the TTM for exercise among individuals living with HIV/AIDS. Furthermore, it is the only known published study that used discriminant analysis to test the predictive ability of the TTM for exercise constructs among individuals living with HIV/AIDS. The overall results suggest that using the TTM to tailor interventions in this population may be better than using indiscriminant mass communication approaches and, as a result, provide support for using the TTM to develop stage-matched interventions among individuals living with HIV/AIDS.
The results of the present study do not constitute endorsement by the American College of Sports Medicine (ACSM).
1. Adams J, White M. Are activity promotion interventions based on the transtheoretical model effective? A critical review. Br J Sports Med
2. Adams J, White M. Why don't stage-based activity promotion interventions work? Health Educ Res
4. Basta TB, Reece M, Wilson MG. The transtheoretical model and exercise among individuals living with HIV. Am J Health Behav
5. Bunton R, Baldwin S, Flynn D, et al. The 'stages of change' model in health promotion: science and ideology. Crit Public Health
6. Burnam A, Bing EG, Morton SC, et al. Use of mental health and substance abuse treatment services among adults with HIV in the United States. Arch Gen Psychiatry
7. Cardinal BJ. Predicting exercise behavior using components of the transtheoretical model of behavior change. J Sport Behav
8. Cardinal BJ. The stages of exercise scale and stages of exercise behavior in female adults. J Sports Med Phys Fitness
9. Cardinal BJ, Kosma M, McCubbin JA. Factors influencing the exercise behavior of adults with physical disabilities. Med Sci Sports Exerc
10. Cardinal BJ, Tuominen KJ, Rintala P. Cross-cultural comparison of American and Finnish college students' exercise behavior using transtheoretical model constructs. Res Q Exerc Sport
11. Cardinal BJ, Keis JS, Ferrand C. Comparison of American and French college students' stage of change for muscular fitness-promoting behaviors. Am J Health Promot
12. Centers for Disease Control and Prevention. Promoting Physical Activity
. Champaign, IL: Human Kinetics. 1999;58-62.
15. Cronbach LJ. Coefficient alpha and the internal and the internal structure of tests. Psychometrika
16. Duggan J, Peterson WS, Schutz M, et al. Use of complementary and alternative therapies in HIV-infected patients. AIDS Patient Care and STDS
17. Fahrenwald NL, Walker SN. Application of the transtheoretical model of behavior change to the physical activity behavior of WIC mothers. Public Health Nurs
18. Hellmann EA. Use of the stages of change in exercise adherence model among older adults with a cardiac diagnosis. J Cardiopulm Rehabil
19. Herrick AB, Stone WJ, Mettler MM. Stages of change, decisional balance, and self-efficacy across four health behaviors in a worksite environment. Am J Health Promot
20. Huberty CJ, Hussein MH. Some problems in reporting use of discriminant analyses. J Exp Educ
21. Huberty CJ, Olejnik S. Applied MANOVA and Discriminant Analysis
. 2nd ed. Hoboken (NJ): Wiley-Interscience; 2006. p. 253-385.
23. Kosma M, Cardinal BJ, McCubbin JA. Predictors of physical activity stage of change among adults with physical disabilities. Am J Health Promot
24. LaPerriere A, Fletcher MH, Antoni NG, et al. Aerobic exercise training in an AIDS risk group. Int J Sports Med
25. MacArthur RD, Levine SD, Birk TJ. Supervised exercise training improves cardiopulmonary fitness in HIV-infected persons. Med Sci Sports Exerc
26. Marcus BH, Rakowski W, Rossi JS. Assessing the motivational readiness and decision making for exercise. Health Psychol
27. Marcus BH, Selby VC, Niaura RS, et al. Self-efficacy and the stages of exercise behavior change. Res Q Exerc Sport
28. Marcus BH, Simkin LR. The stages of exercise behavior. J Sports Med Phys Fitness
29. Marcus BH, Simkin LR, Rossi JS, et al. Longitudinal shifts in employees' stages and processes of exercise behavior change. Am J Health Promot
30. Marcus BH, Emmons KM, Simkin-Silverman LR, et al. Evaluation of motivationally tailored vs. standard self-help physical activity interventions at the workplace. Am J Health Promot
32. Nigg CR, Rossi JS, Norman GL, et al. Structure of decisional balance for exercise adoption. Ann Behav Med
33. Nigg CD, Norman GJ, Rossi JS, et al. Processes of exercise behavior change: redeveloping the scale. Ann Behav Med
34. O'Hea EL, Wood KB, Brantley PJ. The transtheoretical model: gender differences across 3 health behaviors. Am J Health Behav
35. Perkins DO, Stern RA, Golden RN, et al. Mood disorders in HIV infection: prevalence and risk factors in a non-epicenter of the AIDS epidemic. Am J Psychiatry
36. Perna FM, LaPerriere A, Klimas N, et al. Cardiopulmonary and CD4 cell changes in response to exercise training in early symptomatic HIV infection. Med Sci Sports Exerc
37. Prochaska JO, Redding CA, Evers KE. The transtheoretical model and stages of change. In: Glanz K, Rimer BK, Lewis FM, editors. Health Behavior and Health Education
. 3rd ed. San Francisco (CA): Jossey-Bass; 2002. p. 99-120.
38. Rigsby LW, Dishman RK, Jackson AW, et al. Effects of exercise training on men seropositive for the human immunodeficiency virus-1. Med Sci Sports Exerc
39. Rodgers WM, Courneya KS, Bayduza AL. Examination of the transtheoretical model and exercise in 3 populations. Am J Health Behav
40. Sallis JF, Haskell W, Wood P, et al. Physical activity assessment methodology in the Five-City project. Am J Epidemiol
41. Sarkin JA, Johnson SS, Prochaska JO, et al. Applying the transtheoretical model to regular moderate exercise in an overweight population: validation of a stages of change measure. Prev Med
42. Stevens J. Applied Multivariate Statistics for the Social Sciences
. Mahwah, NJ: Erlbaum, Lawrence & Associates, 2002, p. 627.
43. Sullivan A, Nelson M. Marked hyperlipidemia on ritonavir therapy. AIDS
44. Sutton S. Back to the drawing board? A review of applications of the transtheoretical model to substance abuse. Addiction
45. Stringer WW, Berezovskaya M, O'Brien WA, et al. The effect excercise training on aerobic fitness, immune indices, and quality of life in HIV + patients. Med Sci Sports Exerc
46. Tai-Seale T. Stages of change specific triggers and barriers to moderate physical activity. Am J Health Behav
47. Treisman G, Angelino AF, Hutton H. Psychiatric issues in the management of patients with HIV infection. JAMA
48. Wagner G, Rabkin J, Rabkin R. Exercise as a mediator of psychological and nutritional effects of testosterone therapy in HIV+ men. Med Sci Sports Exerc
49. Weinstein ND, Rothman AJ, Sutton SR. Stage theories of health behavior: conceptual and methodological issues. Health Psychol
50. Woods C, Mutrie N, Scott M. Physical activity intervention: a transtheoretical model-based intervention designed to help sedentary young adults become active. Health Educ Res