Cervical cancer screening interventions for U.S. Latinas: a systematic review.
A Cochrane Collaboration systematic review was conducted on interventions to improve the rates of cervical cancer screening among women in general (Forbes, Jepson, & Martin-Hirsch, 2011). Thirty-five studies met the inclusion criteria. The strongest evidence was for invitation letters mailed to women for cervical cancer screens. Fixed appointments were better than invitations with open appointments. Limited evidence was provided for telephone invitations, and it was not clear whether they offered benefits over invitation letters. Overall, education about cervical cancer and its screening was more beneficial than no intervention or usual care, but there was not sufficient evidence to support any one type of education over another. Although Forbes et al. (2011) found some limited evidence for the use of lay community members to provide culturally sensitive information to ethnic minority women, the full range of interventions offered to Latina women to improve PAP screening rates has not been examined through systematic review.
The primary purpose of this systematic review, therefore, was to determine the magnitude and direction of the association between participation in an intervention to increase cancer prevention behavior among Latinas and the cervical screening (that is, PAP) rates of these participants. The secondary purpose was to determine whether certain moderators influenced the effects of these outcomes.
The inclusion criteria for this review involved the following:
* Studies that involved experimental or quasi-experimental designs in which a control (no treatment), intervention as usual, or comparison intervention was used.
* Interventions studies sought to improve cervical cancer screening rates for Latinas.
* Participants of studies were Latina adults ([greater than or equal to] 18 years old) living in the United States; if other ethnic groups were present in studies, results were reported separately for Latinas.
* Both published and unpublished literature were included.
Databases were searched with no restrictor date on the start date of publication until January 2009. Two research librarians developed the search terms of "(latina * or hispanic *) and (PAP or cervical cancer) and (prevent * or intervention* or promot* or encourag*)." Two trained master's level advanced research course students independently searched the following databases with the number of "hits" presented in parentheses: CINAHL (106), Dissertation Abstracts International (25), GenderWatch (112), MEDLINE/ PubMed (167), and PsyclNFO (69). In addition, the master's students conducted a hand search of relevant reference lists of reviews of cervical cancer screening and studies that were included in the systematic review. They also contacted, by e-mail, authors of included studies for their knowledge of other studies.
The first author developed and adapted a data extraction instrument from Littel, Corcoran, and Pillai (2008). Coding sheets for each study included information on sample characteristics, type of intervention, methodological characteristics, and quality of study. Two trained graduate social work students enrolled in an advanced research course coded the studies independently. After they coded the studies, the students compared coding sheets and resolved discrepancies with the assistance of the first author.
The second author performed all statistical analyses with NCSS 2007 statistical and power analysis software (Hintze, 2007), which is a multipurpose statistical package with procedures for conducting meta-analyses and supporting graphs and diagrams. Analyses included the following: study-specific odds ratios (ORs) for screened (success) versus not screened (failure) with corresponding 95 percent confidence intervals (CIs), Cochran's Q statistics, and forest plots.
An OR is defined as the ratio of the odds of an event (for example, being screened for cancer) occurring among intervention group participants compared with the odds of that same event occurring among control group participants. The OR is a way of comparing whether the probability of a certain event is the same for two groups. An OR of one implies that the event is equally likely in both groups; an OR greater than one implies that the event is more likely in the first group (that is, intervention); and an OR less than one implies that the event is less likely in the first group. A 95 percent CI for the log OR is obtained as 1.96 SE on either side of the estimate.
The second author calculated ORs for fixed effects and random effects models. In fixed effects models, the study effects estimate the population effect, with the only error being from the random sampling of participants within the studies. In contrast, random effects models assume that variability between effect sizes emerges from participant-level sampling error and from random differences between studies that are associated with variations in experimental procedures and settings. It has been argued that random effects models more adequately mirror the heterogeneity in behavioral studies and use noninflated alpha levels when the requirement of homogeneity has not been met (Hunter & Schmidt, 2000; Mullen, 1989; Rosenthal, 1984). The second author analyzed heterogeneity by computing Cochran's Q statistic, which has an approximate chi-square distribution with p - 1 degree of freedom, where p is the number of categories within each moderator variables (Hedges & Olkin, 1985). Following Hedges and Pigott (2001), we defined statistical significance as p < .10 rather than the more conventional .05, because of the low power of this test (Hedges & Pigott, 2001).
After screening titles and abstracts, the master's research team read the 33 full-text articles. The main reason for excluding full-text articles was the lack of comparison and control groups in the studies. A list of excluded studies is available from the first author. In all, six studies met inclusion criteria and composed the systematic review. The majority of the studies reviewed for this research were conducted with primarily low socioeconomic status samples in Texas and California. In five studies, the participants were identified as being Mexican American, whereas the remaining study identified participants as Latina. The most common type of study intervention was interpersonal outreach, also known as "promotoras," followed by television media. Study methodology is described in Table 1, and program description is detailed in Tables 2 and 3. Quality of studies is listed in Table 4.
Table 5 is organized by OR size and provides the following: proportions and percentages of intervention and control group respondents who were screened, and ORs for screened (success) in intervention groups versus control groups with corresponding 95 percent CIs. ORs ranged from 1.371 (Fernandez et al., 2005), which indicates a weak positive relationship between the intervention and the outcome, to 0.508 (Jibaja-Weiss, 2005), which indicates a moderate negative relationship. The total fixed effects OR was larger than the OR for the random effects model (0.783 versus 0.778). The CI around the OR for the fixed effects model was narrower than the confidence interval around the OR for the random effects model (0.661 to 0.928 versus 0.576 to 1.049).
Figure 1 is a forest plot of individual study ORs and the total of the combined study ORs. It provides a visual representation of the amount of variation between the results of the studies as well as an estimate of the overall result of all the studies together. The results of component studies are shown as squares centered on the point estimate of the result of each study. Symbols for individual studies (that is, squares) are proportionate to sample size. A horizontal line runs through the square to show its 95 percent CI. The overall estimate from the meta-analysis and its CI are at the bottom, represented as a diamond. The center of the diamond represents the pooled point estimate, and its horizontal line signifies the CI. Significance is achieved at the set level if the diamond is clear of the line of no effect.
Results of the meta-analysis may be biased if the probability of a study being published is dependent on its results. In other words, studies with strong positive findings may be more likely to be published. In an attempt to detect publication bias, the second author explored symmetry in the forest plots. In the absence of publication bias, forest plots should be symmetrical with estimates from larger studies in the center, flanked equally on either side by the less precise estimates. The forest plots would be skewed (that is, asymmetrical) in the presence of a publication bias. The forest plot depicted in Figure 1 suggests symmetry.
Also, to account for the file drawer problem and as a measure of the stability of results, the second author calculated a fail-safe N (Rosenthal, 1984). The fail-safe N is the number of undiscovered or unpublished studies with effect sizes (ESs) of zero that would raise the overall ES above a critical value of p = .05 (Wolf, 1986). An Excel spreadsheet was used to calculate a fail-safe N (DeCoster & Iselin, 2005). This spreadsheet calculates a fail-safe N for Cohen's d. ORs were converted to Cohen's d with a formula suggested by Chinn (2000). According to Chinn (2000), d can be adequately approximated as follows: d=ln (OR) / 1.81. That is, Cohen's d equals the natural log of an OR divided by 1.81. For the studies summarized in Table 1 and Figure 1, the fail-safe N = 1,442.54.
When heterogeneity was examined for the five studies, the value of Cochran's Q at 13.13 was significant (p = .0222, df = 5). Because the test of heterogeneity was statistically significant, more emphasis should therefore be placed on the random effects model.
Because the Q statistic for all reviewed studies combined indicated heterogeneity, a moderator variable analysis of OR was conducted by the second author to identify sources of heterogeneity. Analysis of factors that contributed to the heterogeneity between studies was limited by the small number of studies being analyzed, but sample size and length of follow-up are traditionally considered likely sources of systematic design error. Two approaches were used to determine whether sample size contributed to the heterogeneity of ORs between studies. First, the forest plot for these studies was visually scanned (see Figure 1). Second, sample sizes and ORs were correlated. Table 6 is organized by sample size and provides the following information: percentages for the proportion of respondents who were screened for intervention and control groups, and ORs for screened (success) versus not screened (failure) with corresponding 95 percent CIs. No linear pattern between
sample size and OR was suggested by the forest plot, which is consistent with Pearson's r = -0.030 (one-tailed p = .478) for these same two variables.
The same two approaches were used by the second author to determine whether the length of follow-up time between the intervention and screening contributed to the heterogeneity of ORs between studies. First, the forest plot for these studies was visually scanned (see Figure 2). Second, length of follow-up time and ORs were correlated. Table 7 is organized by length of time to follow-up and provides the following information: percentages for the proportion of respondents who were screened for intervention and control groups, and ORs for screened (success) versus not screened (failure) with corresponding 95 percent CIs. A weak to moderate linear association between time to first follow-up and OR is suggested by the forest plot, which is consistent with Pearson's r = -0.443 (one-tailed p = .222) for these same two variables. That is, the greater the time to first follow-up, the less likely the OR is to suggest a desired outcome (that is, an intervention group participant is screened for cancer).
The results of this systematic review show that, taken as a whole, interventions to improve cervical screening rates among Latina do not seem to be effective. In fact, a slight advantage was conferred to the control conditions, although this could be due to random error. Studies were often flawed in numerous ways (see Table 4). The few studies (N = 6) conducted usually relied on quasi-experimental designs, with one community designated to the experimental condition and another targeted as the control group. At the same time, all but one of the studies relied on a no-intervention control group, so the fact that the control groups often performed better in terms of PAP testing is of concern.
[FIGURE 1 OMITTED]
The Cochrane Collaboration systematic review on interventions for women to encourage cervical cancer screening (Forbes et al., 2011) indicated there was some limited support for the provision of culturally tailored education by lay health care workers to women who were ethnic minorities. In this systematic review, we were unable to test the moderator of type of intervention due to the small number of studies, but many of the studies relied on lay health care providers in this way. In addition, all interventions were designed with culturally sensitive components, such as Spanish-speaking services, personal contact to deliver information, and the use of television media. Furthermore, cultural beliefs and attitudes, such as embarrassment, the belief in fate about cancer, the lack of a preventive focus, and fears of deportation, were addressed in interventions. These are all suggestions that have been made in the qualitative studies that have explored Latina attitudes and perceptions about cervical cancer and its screening (Corcoran & Crowley, in press).
[FIGURE 2 OMITTED]
It must be noted that primary studies did not typically report results separately for Latina immigrants and Latina Americans. Moreover, our small number of studies meant that we would not be able to test for this moderator.
It is interesting to note that a systematic review that was conducted on Latina breast cancer prevention shows better results from intervention to encourage mammography (Corcoran, Dattalo, & Crowley, 2010). It could be that mammograms are more acceptable to Latinas than the PAP, which is a more invasive procedure and is related to sexual activity. The PAP may involve a greater degree of cultural taboo related to the proper conduct of women. It could also be that existing tested interventions, although designed with cultural sensitivity in mind, did not focus on provider behavior and some of the financial and health care access challenges associated with barriers to cervical cancer screening for Latinas (Corcoran & Crowley, in press). In the meta-synthesis, a major category of recommendations to reduce barriers to screening involved offering free or low-cost screening, providing on-site child care, and providing transportation to clinics or, alternatively, having mobile clinics come to the communities. Provider behavior that was important in terms of facilitating screening involved gentle, personal, respectful interactions with a provider who the women had been seeing over time. Female Spanish-speaking providers were in particular demand.
Although social workers should be culturally responsive to Latinas in promoting cervical cancer screening, taking into account the factors that have been discussed, the results of this systematic review indicate that specialized programs to address prevention in this population have shown limited effectiveness over usual care. Programs have tended to target the individual woman. Future studies designed by social workers could center on the macro aspects of the problem to improve cervical cancer prevention among Latinas. Research should also strive for methodological rigor in intervention design, implementation, and analysis.
* Indicates that the study was included in this systematic review.
Chinn, S. (2000). A simple method for converting an odds ratio to effect size for use in meta-analysis. Statistics in Medicine, 19, 3127-3131.
Corcoran, J., & Crowley, M. (in press). Latinas' attitudes about cervical cancer prevention: A meta-synthesis. Journal of Cultural Diversity.
Corcoran, J., Dattalo, P., & Crowley, M. (2010). Interventions to increase mammography rates among U.S. Latinas: A systematic review. Journal of Women's Health, 19, 1281-1288.
DeCoster, J., & Iselin, A. M. (2005). Calculating a fail safe N for Cohen's d. Retrieved from http://www.stat-help. com/spreadsheets/Fail%20Safe%20N.xls
* Fernandez, M. E., Gonzales, A., Tortolero-Luna, G., Partida, S., & Bartholomew, L. K. (2005). Using intervention mapping to develop a breast and cervical cancer screening program for Hispanic farmworkers: Cultivando la salud. Health Promotion Practice, 6, 394-404.
* Fernandez-Esquer, M. E., Espinoza, P., Torres, I., Ramirez, A. G., & McAlister, A. L. (2003). A Su salud: A quasi-experimental study among Mexican American women. American Journal of Health Behavior, 27, 536-545.
Forbes, C., Jepson, R., & Martin-Hirsch, P. (2011). Interventions targeted at women to encourage uptake of cervical screening. Cochrane Database of Systematic Reviews, 5: CD002834.
Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. Orlando, FL: Academic Press.
Hedges, L.V., & Pigott, T. D. (2001). The power of statistical tests in meta-analysis. Psychological Methods, 6, 203-217.
Hintze, J. (2007). NCSS 2007 [Computer software]. Kaysville, UT: NCSS, LLC.
Hunter, J. E., & Schmidt, F. L. (2000). Fixed vs. random effects meta-analysis models: Implications for cumulative research knowledge. International Journal of Selection and Assessment, 8, 275-292.
* Jibaja-Weiss, M. L., Volk, R.J., Smith, Q. W., Holcomb, J. D., & Kingery, P. (2005). Differential effects of messages for breast and cervical cancer screening. Journal of Health Care for the Poor and Underserved, 16, 42-52.
Littel, J., Corcoran, J., & Pillai, V. (2008). Systematic reviews and meta-analysis. New York: Oxford University Press.
McDougall, J. A., Madeleine, M. M., Daling, J. R., & Li, C. (2007). Racial and ethnic disparities in cervical cancer incidence rates in the United States, 1992-2003. Cancer Causes & Control, 18, 1175-1186.
Mullen, B. (1989). Advanced BASIC meta-analysis. Hillsdale, NJ: Lawrence Erlbaum.
* Navarro, A. M., Senn, K. L., McNicholas, L.J., Kaplan, R. M., Roppe, B., & Campo, M. C. (1998). Por la vida model intervention enhances use of cancer screening tests among Latinas. American Journal of Preventive Medicine, 15(1), 32-41.
Owusu, G., Eve, S., Cready, C., Koelln, K., Trevino, F., Urrutia-Rojas, X., et al. (2005). Race and ethnic disparities in cervical cancer screening in a safety-net system. Maternal and Child Health Journal, 9, 285-295.
* Ramirez, A. G., Villarreal, R., McAlister, A., Gallion, K. J., Suarez, L., & Gomez, P. (1999). Advancing the role of participatory communication in the diffusion of cancer screening among Hispanics. Journal of Health Communication, 4, 31-36.
Rosenthal, R. (1984). Meta-analytic procedure for Social research. Beverly Hills, CA: Sage Publications.
Suarez, L., Roche, R. A., Pulley, L., Weiss, N. S., Goldman, D., & Simpson, D. M. (1997). Why a peer intervention program for Mexican-American women failed to modify the secular trend in cancer screening. American Journal of Preventive Medicine, 13, 411-417.
Wolf, F. M. (1986). Meta-analysis: Quantitative methods for research synthesis. Beverly Hills, CA: Sage Publications.
Jacqueline Corcoran, PhD, is professor, and Patrick Dattalo, PhD, is professor, School of Social Work, Virginia Commonwealth University, Richmond. Meghan Crowley, MSW, is bereavement coordinator, Evercare Hospice & Palliative Care, Reston, VA. Address correspondence to Jacqueline Corcoran, School of Social Work, Virginia Commonwealth University, 1001 West Franklin Street, P.O. Box 842027, Richmond, VA 23284-2027; e-mail: email@example.com.
Original manuscript received May 1, 2011
Final revision received November 11, 2011
Accepted January 18, 2012
Advance Access Publication November 21, 2012
Table 1: Design and Sampling Information for Included Studies Design Control Condition (Brief Description of Control) Study and n in Each Group Fernandez et al. (2005) Quasi-experimental Intervention (two communities): n = 132 Control (two neighboring communities; no intervention): n = 111 Fernadez-Esquer, Quasi-experimental Espinoza, Torres, Intervention (one community): n = 53 Ramirez, & McAlister Control (one neighboring community; no (2003) intervention): n = 55 Jibaja-Weiss, Volk, Experimental Smith, Holcomb, & Intervention A (random sample): n = 208 Kingery (2005) Intervention B (random sample): n = 206 Control (random sample; no intervention): n = 204 (a) Navarro et al. (1998) Experimental Intervention (random sample): n = 274 Control (random sample; received community living skills): n = 238 Ramirez et al. (1999) Quasi-experimental Intervention (one community): n = 107 Control (one community; no intervention): n = 104 Suarez et al. (1997) Quasi-experimental Intervention (one community): n = 450 Control (one community; no intervention): n = 473 Study Sampling Recruitment Fernandez et al. (2005) Expanded Programme of Immunization sampling. Quadrants scheme used to identify participants who met eligibility criteria Fernadez-Esquer, Detailed five-stage sampling procedure in Espinoza, Torres, both intervention and comparison Ramirez, & McAlister communities (2003) Jibaja-Weiss, Volk, Recruitment through three health clinics; Smith, Holcomb, & random assignment to group Kingery (2005) Navarro et al. (1998) Recruitment through peer counselors; randomization through counselor groups (half to intervention and half to control) Ramirez et al. (1999) Baseline telephone survey of 1,200 residents in each city; participants selected if met baseline criteria Suarez et al. (1997) Detailed five-stage sampling procedure in both intervention and comparison communities Study Demographic Information on Sample Fernandez et al. (2005) Primarily Mexican American, ages 50 and up, varied socioeconomic status (SES) Fernadez-Esquer, Mexican American women, ages 18 Espinoza, Torres, and up, low SES Ramirez, & McAlister (2003) Jibaja-Weiss, Volk, Mexican American women, age 18 and Smith, Holcomb, & up, low SES Kingery (2005) Navarro et al. (1998) Latinas, ages 40-49, low to medium SES Ramirez et al. (1999) Mexican American women, low SES Suarez et al. (1997) Mexican American women, age 40 and up, low SES (a) For the purposes of this systematic review, the researchers only compared intervention B with the control condition. Table 2: Program Description Name of Study Program Description Fernandez et al. Cultivando Lay health workers implemented (2005) la Salud educational program to individuals in the community. Fernadez-Esquer A Su Salud Mammogram messages were delivered et al. (2003) via local mass media and reinforced by indigenous volunteers. Jibaja-Weiss et al. NA Comparison made between (A) form (2005) letter with cervical cancer information, (B) personalized letter with medical records and cervical cancer information, and (C) no letter. (a) Navarro et al. Por la Vida Lay health workers provided (1998) culturally appropriate education and trainings based on social learning theory. Ramirez et al. NA Mass media campaign, pamphlets, (1999) and community peer educators delivered information about cervical cancer. Suarez et al. A Su Salud Media role models and community (1997) volunteers provided information on breast cancer and mammograms. Communitywide Geographic or Study Duration Location Clinic-based Fernandez et al. Not reported Merced & Communitywide (2005) Watsonville, CA; Eagle Pass, TX; Anthony, NM Fernadez-Esquer 21 months San Antonio, TX; Communitywide et al. (2003) Houston, TX Jibaja-Weiss et al. One time Not reported Clinic-based (2005) Navarro et al. 12 weeks San Diego, CA Communitywide (1998) Ramirez et al. Two years Brownsville, TX; Communitywide (1999) Laredo, TX Suarez et al. Three years El Paso, TX; Communitywide (1997) Houston, TX Note: NA = not applicable. (a) For the purposes of this systematic review, the researchers only compared Intervention B with the control condition. Table 3: Characteristics of Interventions Printed Interpersonal Psychoeducational Study TV Media Mailing Outreach Group Fernandez et al. X (2005) Fernadez-Esquer X X et al. (2003) Jibaja-Weiss et X al. (2005) Navarro et al. X X (1998) Ramirez et al. X X X (1999) Suarez et al. X X (1997) Table 4: Methodological Quality of Included Studies Study Allocation Method Attrition Fernandez, Gonzales, Random allocation unmet Unknown Tortolero-Lung, Partida, & Bartholomew (2005) Fernandez-Esquer et Random allocation unmet High attrition rate al. (2003) Jibaja-Weiss et al. Method of random Not applicable (2005) allocation unclear Navarro et al. (1998) Method of random High attrition rate allocation unclear Ramirez et al. (1999) Random allocation unmet Not applicable Suarez et al. (1997) Random allocation unmet No concern Study Other Potential Bias Fernandez, Gonzales, Allocation concealment, detection Tortolero-Lung, bias, validated outcome measures Partida, & Bartholomew (2005) Fernandez-Esquer et Performance bias al. (2003) Jibaja-Weiss et al. Allocation concealment (2005) Navarro et al. (1998) Allocation concealment, detection bias Ramirez et al. (1999) None found Suarez et al. (1997) Allocation concealment Table 5: Odds Ratios, by Size N with Event / Total N (%) Study Intervention Control Fernandez et al. (2005) 32/132 (24) 21/111 (19) Navarro et al. (1998) (a) 130/199 (65) 99/162 (61) Suarez et al. (1997) 231/450 (51) 268/473 (57) Ramirez et al. (1999) 40/107 (37) 51/104 (49) Fernandez-Esquer et al. (2003) 33/111 (30) 37/84 (44) Jibaja-Weiss et al. (2005) 49/208 (24) 77/204 (38) Total (fixed effects) Total (random effects) Study Odds Ratio (95% CI) Fernandez et al. (2005) 1.371 (0.738-2.549) Navarro et al. (1998) (a) 1.199 (0.780-1.843) Suarez et la. (1997) 0.807 (0.623-1.046) Ramirez et al. (1999) 0.620 (0.358-1.074) Fernandez-Esquer et al. (2003) 0.537 (0.297-0.972) Jibaja-Weiss et al. (2005) 0.508 (0.331-0.779) Total (fixed effects) 0.783 (0.661-0.928) Total (random effects) 0.778 (0.576-1.049) Note: CI = confidence interval. (a) Multiple percentages were reported; percentages at first-follow- up were used to provide consistency with other studies in this analysis. Table 6: Odds Ratio, by Sample Size (Random Effects Model) N with Event / Total N (%) Study Intervention Control Suarez et al. (1997) 231/450 (51) 268/473 (57) Jibaja-Weiss et al. (2005) 49/208 (24) 77/204 (38) Navarro et al. (1998) 130/199 (65) 99/162 (61) Fernandez et al. (2005) 32/132 (24) 21/111 (19) Fernandez-Esquer et al. (2003) 33/111 (30) 37/84 (44) Ramirez et al. (1999) 40/107 (37) 51/104 (49) Combined average Study Odds Ratio (95% CI) N Suarez et al. (1997) 0.807 (0.623 to 1.046) 923 Jibaja-Weiss et al. (2005) 0.508 (0.331 to 0.779) 412 Navarro et al. (1998) 1.199 (0.780 to 1.843) 361 Fernandez et al. (2005) 1.371 (0.738 to 2.549) 243 Fernandez-Esquer et al. (2003) 0.537 (0.297 to 0.972) 195 Ramirez et al. (1999) 0.620 (0.358 to 1.074) 211 Combined average 0.779 (0.579 to 1.048) Note: CI = confidence interval. Table 7: Odds Ratio, by Time to First Follow-up (Random Effects Model) N with Event / Total N (%) Study Intervention Control Suarez et al. (1997) 231/450 (51) 268/473 (57) Fernandez-Esquer et al. (2003) 33/111 (30) 37/84 (44) Ramirez et al. (1999) 40/107 (37) 51/104 (49) Fernandez et al. (2005) 32/132 (24) 21/111 (19) Jibaja-Weiss et al. (2005) 49/208 (24) 77/204 (38) Navarro et al. (1998) 130/199 (65) 99/162 (61) Combined average Time to First Odds Ratio Follow-up Study (95% CI) (Months) Suarez et al. (1997) 0.807 (0.623 to 1.046) 48 Fernandez-Esquer et al. (2003) 0.537 (0.297 to 0.972) 30 Ramirez et al. (1999) 0.620 (0.358 to 1.074) 24 Fernandez et al. (2005) 1.371 (0.738 to 2.549) 12 Jibaja-Weiss et al. (2005) 0.508 (0.331 to 0.779) 12 Navarro et al. (1998) 1.199 (0.780 to 1.843) 0 Combined average 0.779 (0.579 to 1.048) Note: CI = confidence interval.
|Printer friendly Cite/link Email Feedback|
|Author:||Corcoran, Jacqueline; Dattalo, Patrick; Crowley, Meghan|
|Publication:||Health and Social Work|
|Date:||Nov 1, 2012|
|Previous Article:||Medicaid and the affordable care act after the supreme court decision.|
|Next Article:||Consumer perspectives on involving family and significant others in a healthy lifestyle intervention.|