![]() |
||||||||
|
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
||
![]() ![]() |
![]() |
![]() |
![]() ![]() |
![]() |
![]()
Volume 8: No. 1, January 2011
ORIGINAL RESEARCH |
Box. Measures for Blood Pressure, Cholesterol, and Glucose Levels; Diabetes Risk; and Dental Score, Mobile Health Screening Intervention Using Resident Health Advocates, Boston, 2007-2008 | |
---|---|
Blood pressure, systolic/diastolic (mm Hg)a | |
Normal | <120/<80 |
Prehypertension | 120-139 or 80-89 |
Hypertension, Stage 1b | 140-159 or 90-99 |
Hypertension, Stage ≥2 b | ≥160 or ≥100 |
Cholesterol (mg/dL)a | |
Normal | <200 |
Borderlineb | 200-240 |
Highb | >240 |
Glucose level (mg/dL)c | |
Normal | ≤140 |
Highb | >140 |
Diabetes risk scorec | |
0-2 | very low risk |
3-9 | low to medium risk |
≥10b | high risk |
Dental scored | |
0 | no obvious problems |
1 | no referral, nonurgent |
2b | referral, nonurgent |
3b | urgent dental care within 24 hrs |
a US Preventive Services Task Force (17). |
In 2007, residents with screen-positive results were offered help in making an appointment at the health center of their choice at both intervention and control sites. At both intervention and control sites, RHAs made appointments either on-site or at a later time, in which case the person being referred was called with the appointment information.
On the basis of our experience with referrals in 2007, we pilot-tested a process in 2008 that involved several steps to improve appointment-making and to ensure that appointments were kept. First, we reduced to 2 pages the consent form requesting participants’ permission for research staff to seek follow-up information. The revised medical intake form also included 2 new boxes to record whether the participant needed a referral and if the appointment was urgent. Van visits were scheduled for the time of day when health centers were more accessible for appointment-making (usually midday). RHAs also accompanied people to their appointments, translated as needed, and offered information on services for the uninsured.
We entered data on demographics, access to health care, health information, and screening results from the mobile health unit intake forms and referral forms into an Excel 2003 spreadsheet (Microsoft Corporation, Redmond, Washington) and conducted statistical analysis with SAS version 9.1 (SAS Institute, Inc, Cary, North Carolina). The primary analysis compared use of the mobile health screening service in the intervention with control conditions. The 2 control housing developments in 2007 constituted the control condition; the intervention condition included the experience at the intervention sites in 2007 in addition to the 2008 experience at the housing developments that had been control sites in 2007. The use of the mobile health screening service was measured as the proportion of the total adult population (aged ≥18) who received screening services. From BHA records, we obtained the number of adult residents by sex, age, race/ethnicity, and primary language spoken at home.
We calculated relative risk (RR) for participants’ attendance with 95% confidence intervals (CIs) to compare the proportion of intervention and control condition residents who attended screenings. We calculated P values by using χ2 or t tests as appropriate. Because there were no morning screenings in 2008 and, therefore, more afternoon and evening visits in the intervention condition, we calculated a standardized RR, weighted by the total number of morning (n = 4), afternoon (n = 8), and evening (n = 6) sessions in 2007 and 2008. We also compared the distribution of participants at intervention and control sites in 2007 by sex, age, race/ethnicity, time of screening (morning, afternoon, or evening), primary language spoken at home, education, medical and dental insurance status, and most recent primary care visit.
To assess the role of RHAs in facilitating follow-up medical care after a positive screening, we compared the proportion of consent forms completed (to gauge the success of the enrollment process) and the proportions of appointments made and kept in the summer of 2007 with those in the summer of 2008. We combined intervention and control sites in 2007 because there were no differences in the nature or intensity of efforts to increase follow-up at either set of sites.
In 2007, 6% (n = 100) of adult residents at the 2 intervention sites were screened, compared with 3% (n = 47) of residents from control sites (RR, 1.74; 95% CI, 1.24-2.44). Use at the intervention sites, adjusted for time of day of mobile health screening visits, was also higher than at control sites for both years combined (RR, 1.55; 95% CI, 1.12-2.15).
For both intervention and control sites combined in 2007, mobile health screening participants were primarily female, Hispanic, had a mean age of 44, and had completed high school (Table 2). Hispanic (57% of mobile health screening participants compared with 41% of adult residents, P < .001) and male residents (39% of mobile health screening participants compared with 27% of adult residents, P = .001) used the van at rates disproportionate to their numbers in the developments.
Of the 224 participants from intervention and control sites across both years, 146 (65%) had at least 1 positive screening diagnosis for hypertension, high cholesterol, diabetes risk, or dental disease. Of the 217 participants screened for hypertension, 64 had either stage 1 (n = 36) or stage 2 (n = 28) hypertension. Twenty of the 64 participants with stage 1 or stage 2 hypertension had not seen a doctor in the past 12 months, and 39 were considered an untreated positive screen (Table 3).
Although 25% of participants had a screening diagnosis of hypertension and 24% had high cholesterol, findings of diabetes risk and dental disease were more common. Two-thirds of those screened (n = 114) had a high diabetes risk score based on the 8-question self-administered tool, and 9 of the 114 had a positive blood glucose test (Table 3). A total of 127 people attended the mobile health screening service on days when screening for dental disease was provided, 49 of whom chose not to use dental services. Of the 78 who were screened, 41 (53%) were referred to a dentist for follow-up. Referrals were based on a 0-3 scoring system used by the dental staff (Box). Urgent care was recommended for 3 participants who reported significant pain or had obvious infection.
In 2007, among the 91 participants who screened positive for any condition, 44 (48%) provided consent for follow-up. Of these, appointments were made for 27 participants (61%) within 3 months after the date of the screen-positive finding. Appointments were not made for 17 participants for 2 major reasons: disconnected telephone lines or inability to contact the resident after several telephone calls (n = 8) and system barriers (n = 9). Fourteen (52%) of the 27 participants with appointments kept them. In 2008, among the 44 participants who screened positive, 33 (75%) provided consent for follow-up. Of those who consented, appointments were made for all 33 participants within 1 month after the date of the screen-positive finding. Of the 33 participants with appointments, 24 (73%) had kept an appointment within 2 months after the initial screening. Overall, the proportion of all screen-positive participants who completed a follow-up medical appointment through the mobile health screening mechanism increased from 15% (14 of 91) in 2007 to 55% (24 of 44) in 2008.
Community-based interventions are most successful when they are designed with communities as respected partners, address problems in the context of community strengths, respect cultural diversity, and use outreach workers (20-22). Three previous studies have reported success in the use of RHAs in low-income and senior citizen housing to boost health promotion practices related to smoking cessation (10), prevention of HIV infection (21,23), and mammography (24).
In our study, we recruited and trained public housing residents as RHAs to motivate other public housing residents to use a city-run mobile public health unit for health screening. The results of the intervention showed that intervention sites with RHAs had higher rates of screening for chronic diseases on the mobile health screening service compared with control sites with no RHAs. Although we cannot isolate individual recruitment strategies that may have been the most responsible, RHAs actively and frequently distributed colorful, bilingual flyers at intervention sites; recruited fellow residents as peer leaders; provided one-to-one motivational advice; and were accessible to fellow residents.
In previous studies, outreach workers were associated with significant increases in cervical cancer screening (14), mammography education (24), and diabetes education (12). The magnitudes of these increases were 17%, 40%, and 70%, respectively. Such results are similar in magnitude to the results obtained in this study, in which medical visits increased to 55% (a relative difference of 72%) among populations receiving the RHA-delivered screening promotion message.
Although the RHA intervention increased use of on-site van screening services, participants represented only about 6% of adults at the intervention developments. On average, about 5 people per hour were screened, well below the potential capacity of 10 to 12 visits per hour. However, our intervention was limited to one 4-hour visit per month to each development for 3 successive months. We believe that additional visits would have increased use and that continued presence of the van would have resulted in van use by a substantial number of residents.
During the first summer, only 15% of residents with screen-positive findings had a follow-up medical appointment. We identified important barriers and, before the next summer, we developed training programs, an expedited referral process, and new intake forms. We scheduled van visits to coincide with better calling hours to health centers and added a component in which RHAs accompanied residents to follow-up appointments. As a result, compared with the first summer, rates of appointments kept increased more than threefold. Furthermore, all 3 individual components (consent provided, appointments made, and appointments kept) contributed to the overall increase.
This study has a number of limitations. Although the intervention and control sites in 2007 were pair-matched on the basis of location, size, and accessibility to a community health center, each development maintains a unique set of characteristics that makes comparison of sites difficult. Second, the personality and social networks of the individual RHAs may have been an important factor in the successful dissemination of the screening message, but we could not assess these variables. Third, we did not track subsequent use of medical care among attendees who had a screen-positive result if they did not consent or were lost to follow-up in the first year. As a result, we may have overestimated the effect of RHA referral activities. However, the size of the increase in completed appointments from the first to second year was so large that some improvement can probably be attributed to the efforts of the RHAs. Finally, we did not conduct a systematic evaluation of the reasons that residents chose not to participate in the screening, so we do not know the number of residents who should be considered in the target population for screening services nor what barriers prevented their participation.
In conclusion, we found that RHAs significantly increased health screening among residents of Boston public housing developments. In addition, we found high levels of 4 chronic conditions among Boston public housing residents. These high rates underscore the need for expanded screening services, enhanced access to primary care providers, and improved referral networks to treat chronic disease. High rates of self-assessed risk indicate potential benefits of preventive programs for diabetes that use expanded nutritional and exercise counseling. RHA recruitment of fellow residents for screening, if sustained and coupled with clinical follow-up and adherence to medical recommendations, would improve health and would reduce health disparities among public housing residents. Further research is needed to assess whether increases in screening translate into increased clinical follow-up and participation in health-promotion programs.
We thank Lisa Cooper of the Centers for Disease Control and Prevention, Gerry Thomas of the Boston Public Health Commission, Greg Davis of the Boston Housing Authority, resident health advocates Juanely Bautista, Berkis Diez, Ramona Lara, Maria Patron, Amanda Reira, Sharon Wilson, and Eddie Woyman, and Kathy M. Lituri, RDH, MPH, oral health promotion coordinator, Boston University Goldman School of Dental Medicine Division of Community Health Programs.
Corresponding Author: Jo-Anna L. Rorie, CNM, MSN, MPH; Department of Community Health Sciences, Boston University School of Public Health, 801 Massachusetts Ave #442, Boston, MA 02118-2526. Telephone: 617-638-5058. E-mail: jrorie@bu.edu.
Author Affiliations: Adriana Smith, Tegan Evans, C. Robert Horsburgh Jr, Daniel R. Brooks, Rachel Goodman, Doris Bunte, Lee Strunin, Daisy de la Rosa, Alan Geller, Boston University Goldman School of Dental Medicine, Division of Community Health Programs, Boston Massachusetts.
![]() |
![]() |
![]() Tables
|
Characteristic | Study Site | |||
---|---|---|---|---|
Intervention A,b No. (%) | Intervention B, No. (%) | Control A, No. (%) | Control B, No. (%) | |
Age and sex | ||||
Residents aged ≥18 years | 1,181 (100) | 534 (100) | 1,045 (100) | 354 (100) |
Women | 814 (69) | 416 (78) | 773 (74) | 272 (77) |
Race/ethnicity | ||||
Black | 283 (24) | 250 (47) | 522 (50) | 152 (43) |
Hispanic | 401 (34) | 267 (50) | 459 (44) | 159 (45) |
White | 330 (28) | 16 (3) | 31 (3) | 21 (6) |
Language spoken by head of household | ||||
Primary language Spanish | 389 (33) | 240 (45) | 438 (42) | 145 (41) |
Primary language English | 590 (50) | 245 (46) | 491 (47) | 187 (53) |
Proximity and mobile health screening service history | ||||
Closest community health center | On-site | <1 mile | On-site | <1 mile |
Mobile health screening history | New site | Old site | New site | Old site |
a Numbers may not total the number of residents
and percentages may not total 100 because missing data were not counted.
b Intervention sites A and B were pair-matched with control sites A and B, respectively.
Characteristic | Intervention | Control | P Valueb | ||
---|---|---|---|---|---|
Participants, No. (%), n = 100 | Population ≥18 y, No. (%), n = 1,715 | Participants, No. (%), n = 47 | Population ≥18 years, No. (%), n = 1,399 | ||
Age, mean (SD), y | 45.1 (18.7) | NA | 42.8 (17.2) | NA | .47 |
Sex | |||||
Women | 60 (60) | 1,235 (72) | 29 (62) | 1,049 (75) | .76 |
Men | 40 (40) | 480 (28) | 18 (38) | 350 (25) | |
Race/ethnicity | |||||
Black | 22 (22) | 497 (29) | 17 (36) | 672 (48) | .01 |
Hispanic | 56 (56) | 669 (39) | 28 (60) | 616 (44) | |
White | 15 (15) | 326 (19) | 0 | 56 (4) | |
Other | 6 (6) | 223 (13) | 2 (4) | 55 (4) | |
Languagec | |||||
English | 35 (35) | 840 (49) | 18 (38) | 686 (49) | .92 |
Spanish | 57 (57) | 617 (36) | 26 (55) | 588 (42) | |
Haitian Creole | 4 (4) | NA | 1 (2) | NA | |
Other/unknown | 4 (4) | 258 (15) | 2 (4) | 125 (9) | |
Education | |||||
Did not attend school | 2 (2) | NA | 4 (6) | NA | .52 |
Some primary or secondary | 40 (40) | NA | 19 (40) | NA | |
High school graduate | 29 (29) | NA | 14 (30) | NA | |
Post high school education | 29 (29) | NA | 10 (21) | NA | |
Health insurance | |||||
None | 30 (30) | NA | 10 (21) | NA | .46 |
State-provided | 59 (59) | NA | 29 (62) | NA | |
Private | 11 (11) | NA | 8 (17) | NA | |
Dental insurance | |||||
Yes | 57 (57) | NA | 28 (60) | NA | .85 |
No | 43 (43) | NA | 19 (40) | NA | |
Last visit to doctor | |||||
<6 months ago | 52 (52) | NA | 26 (55) | NA | .55 |
6-11 months ago | 22 (22) | NA | 6 (13) | NA | |
1-2 years ago | 16 (16) | NA | 8 (17) | NA | |
>2 years ago | 11 (11) | NA | 7 (15) | NA |
Abbreviation: NA, not available or not assessed.
a Percentages may not total 100%
because of rounding.
b P value represents comparison between
participants at intervention and control sites.
c Language for mobile health participants represents participant’s first language; language at sites represents household language spoken.
Chronic Condition | No. With Positive Screening Results | No. Not Previously Aware of Their Conditiona | No. With No Recent Medical Careb |
---|---|---|---|
Hypertension: 217 total screenings | |||
Stage 1 | 36 | 22 | 10 |
Stage 2 | 28 | 17 | 10 |
Cholesterol: 200 total screenings | |||
Borderline | 35 | 27 | 6 |
High | 12 | 10 | 5 |
Diabetes: 172 risk testsc, 103 glucose screenings | |||
High riskd | 114 | 88 | 30 |
High glucose | 9 | 6 | 2 |
Dental care: 78 total screenings | |||
Nonurgent care | 41 | 4 | 13 |
a An “untreated positive screen” is a screen-positive result in a participant who was not previously aware of having the condition
(undetected) or who had not been taking a prescribed medication for the condition
(detected but untreated).
b Participants who had positive screening results for
both hypertension and diabetes
and had not been seen by a physician in >12 months.
c To assess diabetes risk, we used a standard 8-question self-administered tool that asked about diet, exercise, age, and diabetes in the family (18).
d Participants with a score of 10 or more on the diabetes risk test were considered at high risk and were offered a blood glucose screening test.
Comment on this
article at PCD Dialogue
Learn more about PCD's
commenting policy
![]() |
![]() |
![]() |
|
![]() |
![]() |
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention. ![]()
Privacy Policy | Accessibility This page last reviewed March 30, 2012
|
|