222-D4

RESEARCH AND PRACTICE

Results of the 2004 National Worksite Health Promotion Survey I Laura Linnan, ScD, CHES, Mike Bowling, PhD, Jennifer Chiidress, MS, CHES, Garry Lindsay, MPH, CHES, Carter Blakey, Stephanie Pronk, MEd,

Sharon Wieker, and Penelope Royall, MSW, PT

Worksites are important public health settings because the majority of US adults spend con- siderable amounts of time at work, and the work environment exerts an independent in- fluence on employee health. Addressing both the work environment and individual health behavior is essential to producing gains in employee health.'"^ In addition, the "health" of a business depends on strategies that man- age both business costs and employee health care costs. Thus, tracking employer efforts to promote health is warranted.

In the United States, the first national worksite health promotion survey was con- ducted in 1985, and follow-up surveys were conducted in 1992, 1999, and 2004. These surveys serve as national benchmarks and as indicators of change over time. One major worksite health-related goal included in Healthy People 2010 is to increase to at least 7 5 % the number of employers that offer a comprehensive health promotion program for employees."*'' We examined data form the 2004 National Worksite Health Promotion Survey to monitor the prevalence of worksite health promotion programs, policies, services, and supportive environments and to assess the implications of the survey's results for public health practice and research.

METHODS

Study Design and Sample The 2004 National Worksite Health Pro-

motion Survey gathered information from a cross-sectional, nationally representative sam- ple of US worksites. The sample was drawn from the Dun & Bradstreet database^ of all private and public employers in the continen- tal United States. To the extent possible, the survey's procedures followed those used in previous national surveys''* so that between- survey comparisons could be made.

Questions addressed spécifie worksites rather than the companies to which the worksites belonged. The survey involved a disproportionate stratified sampling design

Objectives. We examined worksite health promotion programs, policies, and services to monitor the achievement of the Healthy People 2010 worksite-related goal of 75% of worksites offering a comprehensive worksite health promotion program.

Methods. We conducted a nationally representative, cross-sectional telephone survey of worksite health promotion programs stratified by worksite size and in- dustry type. Techniques appropriate for analyzing complex surveys were used to compute point estimates, confidence intervals, and multivariate statistics.

Results. Worksites with more than 750 employees consistently offered more programs, policies, and services than did smaller worksites. Only 6.9% of re- sponding worksites offered a comprehensive worksite health promotion pro- gram. Sites with a staff person dedicated to and responsible for health promo- tion were significantly more likely to offer a comprehensive program, and sites in the agriculture and mining or financial services sector were significantly less likely than those in other industry sectors to offer such a program.

Conclusions, Increasing the number, quality, and types of health promotion pro- grams at worksites, especially smaller worksites, remains an important public health goal. {Am J Public Health. 2008:98:1503-1509. doi:10.2105/AJPH,2006.100313)

with 35 strata defined according to 2 cate- gories: number of employees (fewer than 50, 5 0 - 9 9 , 100-249, 250-749, 750 or more) and US Standard Industrial Classification code (agriculture/mining/construction, finance/insurance/real estate, transportation/ communications/utilities, business/professional services, manufacturing, wholesale/retail trade, public administration/government).

Because of the preponderance of work- sites with fewer than 50 employees, we oversampled sites with more than 50 em- ployees to ensure that estimates would be appropriate for all sites of all sizes. We re- port results only for nongovernmental work- sites with 50 or more employees because (1) point estimate variances were unstable for sites with fewer than 50 employees and (2) previous national surveys omitted gov- ernment workplaces.

Data Collection Procedures Trained interviewers conducted the 2004

survey by telephone (each interview required approximately 20 minutes). At each worksite, respondents were identified as being "directly responsible for health promotion or Wellness" or as having an "in-depth knowledge of these types of programs at the worksite."

Response rates were enhanced via several techniques. For example, respondents were provided with a fact sheet describing the im- portance of partidpating in the survey, an- swers to typically asked questions, and a toll- free telephone Une to establish a convenient interview time. Also, interviewers were pro- vided access to a telephone number lookup service to assist in contacting employers that did not answer after 5 call attempts. Finally, standardized guidelines^ were used to recon- tact sites initially unwilling to take part to en- list their partidpation.

Measures Key measures included worksite size (total

number of full- and part-time employees), in- dustry type (Standard Industrial Classification code), number of years the worksite had of- fered a health promotion program (labeled "experience"), and barriers to offering a health promotion program. "Comprehensive" health promotion programs were defined as those that incorporated all of the 5 key ele- ments outlined in Healthy People 2010: (1) health education (i.e., skill development and lifestyle behavior change, along with in- formation dissemination and awareness build- ing), (2) supportive sodal and physical work

August 2008, Vol 98, No. 8 | American Journal of Public Health Linnan et al. \ Peer Reviewed | Research and Practice | 1503

RESEARCH AND PRACTICE

environment (i.e., support of healthy behav- iors and implementation of policies promoting health and reducing risk of disease), (3) inte- gration (i.e., integration of the program into the organization's structure), (4) linkage (i.e., linkage to related programs such as employee assistance programs), eind (5) worksite screen- ing and educaton (i.e., programs linked to ap- propriate medical care).

Data Analysis All analyses were carried out with the

SURVEYFREQ and SURVEYLOGISTIC pro- cedures in SAS/STAT,'" in which Taylor ex- pansion approximations are used to calculate standard errors and their corresponding 9 5 % confidence intervals for stratified weighted data." Weights were computed as the inverse of selection probabilities and were adjusted for nonresponse. Twenty-three worksites from the 7 strata representing sites with fewer than 50 employees were misclassified and thus were reallocated and weighted to the correct size strata. Analyses excluding and including these 23 reclassified sites yielded identical results.

We calculated point estimates with 9 5 % confidence intervals for all of the measures examined and used the Rao—Scott x^ statistic to assess differences according to size and in- dustry type." The level of significance was set at a < . 0 5 . We used the Wald x^ statistic to compare logistic regression models fit to groups with and without a comprehensive health promotion program.'^

RESULTS

Sample Description We conducted a total of 1553 interviews

with worksites from the different size and in- dustry categories. Respondents were weighted across industry and size categories, and sam- ple distributions reflected those in the popula- tion of all eligible worksites. The overall re- sponse rate (corresponding to response rate 3 of the American Association for Public Opin- ion Research'^ guidelines) was 59.7%.

The sample size was 730, excluding govern- mental worksites and those with fewer than 50 employees. The site size breakdown was as follows; 179 sites with 50 to 99 employees, 229 sites with 100 to 249 employees, 211

sites with 250 to 749 employees, and 111 sites with more than 750 employees. Industry categories represented were manufacturing (n= 198), finance (n=85), wholesale or retail (n=117), transportation (n=73), agriculture (n=86), and business or professional (n= 171).

Most survey respondents were directors or managers (60.5%) and were members of ei- ther a human resources or benefits depart- ment (52.7%). Approximately 3 9 % of re- spondents reported a 10% to 15% increase in health care costs in recent years; 31.0% reported an increase of less than 10%, 18.7% reported an increase of more than 20%, and 8.5% reported an increase of 15% to 20%. Overall, 2.5% indicated that they did not offer health care benefits.

Staffing, Experience, and Funding The majority of worksites (64.6%) em-

ployed at least 1 full- or part-time staff person who was directly responsible for health pro- motion and worksite Wellness. Of the sites with health promotion programs, 60.8% indi- cated that their program had been in place for 5 years or less,. 8.7%, for 6 to 9 years, and 30.5%, for 10 or more years. The health plan was identified as the leading source of funding for programs (e.g., health screenings, health risk appraisals, disease management) other than health awareness and information programs, which were most Irequently funded by the employer (47.7%). In all cases, 2% or fewer of responding worksites identi- fied employees or outside vendors as primary sources of funding.

Approximately 2 6 % of worksites reported using incentives to increase employee partici- pation. Incentives involving gifts and dis- counts were mentioned most often, followed by cash incentives. The 48 sites that offered cash incentives reported that the mean amount offered (before taxes) per person per year was $556.88 (SD=$176.70). No differ- ences according to worksite size or industry type were observed in regard to use of incentives.

Evaluation Metiiods, Support, and Barriers to Success

When asked about methods used in pro- gram delivery, respondents most Irequentiy reported using printed materials, followed by

the Internet, in-person strategies, and the tele- phone. For example, in the health awareness and information programming category, 46.0% of sites reported using printed materi- als, 28.1% reported using the Internet, 24.4% reported using in-person methods, and 11.4% reported using telephone ap- proaches. This pattern was consistent in the different program categories with the excep- tion of health risk appraisals; in this category, an identical percentage of respondents (11.0%) reported use of print materials and in-person strategies, whereas 7.8% reported use of the Internet, and 6.4% reported use of telephone approaches.

Approximately 70% of respondents indi- cated that their health promotion program supported the organization's business strat- egy, 67.5% believed that the program was in- tegrated into the overall strategy the em- ployer used to address health care, and 66.2% reported that it was linked to other key organizational areas. However, only 49.5% of sites used data to guide program di- rection, and only 30.2% had a 3- to 5-year strategic plan in place for worksite health promotion.

The most commonly reported barriers or challenges to the success of health promotion programs were lack of employee interest (63.5%), staff resources (50.1%), funding (48.2%), participation on the part of high-risk employees (48.0%), and management sup- port (37.0%). No differences in barriers were reported on the basis of industry type or worksite size, with the single exception that worksites with more than 750 employees were significantly more likely than were smaller sites to report lack of participation among high-risk employees (P=.OO2).

Overall, 19.4% of worksites reported using health risk appraisals, and there were statistically significant differences according to worksite size. For example, only 11.3% of sites with 50 to 99 employees used health risk appraisals, as compared with 45.8% of sites with more than 750 employees (P<.001). When asked about what they used to evaluate program success, respon- dents most often cited employee feedback (73.2%), employee participation (57.4%), workers' compensation costs (57.1%), health care claims costs (57.0%), and time lost or

1504 I Research and Practice | Peer Reviewed | Linnan et ai. American Journai of Pubiic iHeaith | August 2008, Vol 98, No. 8

RESEARCH AND PRACTICE

absenteeism (43.9%). Approximately 4 4 . 1 %

of sites expected a return on investment for

their program; of these sites, 36.2% ex-

pected a return on investment within 12 to

17 months, 23.9% expected it within 18 to

23 months, and 13.4% expected it in less

than 12 months.

Programs, Activities, Screenings, and

Disease Management

The most common types of programs of-

fered were employee assistimce programs

(progrsuns typically offering mental health or

counseling services; 44.7%), followed by

back injury prevention programs or activities

(45%), stress management programs (24.9%),

nutrition programs (22.7%), health care con-

sumerism programs (21.6%), and weight

management programs (21.4%). There was a

clear dose-response relationship in that work-

sites vÁÜi more employees offered more pro-

grams, classes, and activities (Table 1). The

only exception to this pattern was HIV/AIDS

education and health care consumerism; no

differences by worksite size reached statistical

significance.

Respondents were asked whether, in the

past 12 months, they had offered hesilth

screenings to their employees and their fami-

lies through the worksite, health plan, or both.

Blood pressure screenings were most fre-

quentiy offered (36.4%), followed by alcohol

or drug abuse support (35.9%), blood choles-

terol screenings (29.4%), diabetes screenings

(274%), and cancer screenings (21.8%).

Again, sites with more employees consistently

offered more screening services. Between

70% and 8 5 % of worksites with more than

750 employees reported offering all of these

services (blood pressure screening, 84.9%;

blood cholesterol screening, 80.5%; alcohol

or drug abuse support, 70.7%; cancer screen-

ing, 70.2%; and diabetes screening, 70.2%)

(Table 1).

In terms of disease management programs,

2 6 . 1 % of sites offered cardiovascular disease

programs, 2 5 % offered diabetes programs,

16.4% offered obesity programs, and 15.6%

TABLE 1-Selected Health Promotion Programs and Services, t>y Worksite Size: Natlonai Worksite Heaith Promotion Survey, 2004

Programs or activities

Employee assistance

Smoking cessation

Physicai activity

Choiesteroi reduction

Nutrition

Stress management

Weight management

Back Injuiy prevention

Heaith care consumerism'

HiV/AiDS'

Screenings or counseiing services

Cancer screening

Diabetes screening

Biood pressure screening

Biood choiesteroi screening

Aicohol or drug abuse support

Disease management programs

Diabetes

Asthnia°

Cancer'

Depression'

Hypertension'

Back pain'

Cardiovascular disease

Chronic obstructive puimonaiy disease'

Obesity

High-risk pregnancy

Total (n-730), % (95% Ci)

44.7(39.28,50.13)

18.6(14.51,22.46)

19.6(15.54,23.67)

19.9(15.55,24.14)

22.7(18.16,27.24)

24.9(20.10,29.86)

21.4(16.94,25.93)

45.0(39.28,50.65)

21.6(16.76,26.48)

14.6 (10.53,18.70)

21.8(17.45,26.09)

27.4(22.47,32.25)

36.4(30.98,41.74)

29.4(24.50,34.39)

35.9(30.76,41.09)

25.0(20.10,29.83)

19.1(14.84,23.39)

22.5(17.66,27.28)

20.5(16.11,24.87)

22.9(18.10,27.60)

20.1(15.59,24.57)

26.1(21.14,31.10)

15.6 (11.62,19.61)

16.4(12.22,20.53)

18.6(14.22,22.94)

50-99 Employees (n = 179), %(95%a)

32.4(23.49,41.28)

8.8 (3.51,14.12)

9.0 (3.67,14.30)

16.4(9.02,23.87)

11.0(4.61,17.34)

17.6(9.92,25.19)

11.3 (5.11,17.40)

37.2(27.70,46.67)'

16.5(8.64,24.34)

11.3 (4.55,18.12)

14.3(7.82,20.74)

19.0(11.50,26.56)

27.1(18.22,35.92)

21.8(13.77,29.91)

28.6(20.14,37.03)

21.8(13.45,30.08)

15.8(8.64,22.95)

17.5(9.61,25.44)

15.5(8.44,22.64)

20.1(11.87,28.31)

16.1(8.71,23.42)

20.1(12.73,29.22)

13.3 (6.59,19.98)

11.9 (5.12,18.61)

14.8(7.39,22.14)

100-249 Empioyees(n-229), %(95%Ci)

48.07(39.03,57.12)

19.4(12.66,26.08)

23.6(16.11,31.11)

17.5(11.41,23.55)

30.4(21.92,38.85)

27.7(19.44,35.92)

24.8(16.79,32.86)

46.1(37.08,55.11)

27.0 (18.59,35.35)

14.2(7.54,20.92)

22.1(14.90,29.27)

27.7(19.67,35.68)

35.8(27.15,44.35)

26.8(19.13,34.49)

37.3(28.96,45.65)

22.4(15.40,29.39)

20.8(13.97,27.65)

25.8(17.78,33.74)

24.3(16.88,31.69)

23.3(15.94,30.72)

22.3(14.86,29.72)

27.8(20.04,35.59)

14.3(8.55,20.05)

16.8(10.00,23.56)

18.8(12.35,25.21)

250-749 Empioyees(n-211), % (95% Cl)

63.3(52.40,74.24)

32.0(21.92,42.17)

28.5(19.50,37.42)

29.3(19.78,38.86)

34.0(23.50,44.45)

32.3(22.20,42.49)

34.1(23.81,44.43)

55.7(44.88,66.56)

22.7(14.69,30.69)

24.9(15.51,34,38)

29.4(20.06,38.67)

39.9(29.39,50.32)

51.5(40.41,62.69)

43.5(32.94,54.20)

45.0(34.20,55.78)

33.6(23.67,43.53)

18.7(12.08,25.37)

27.9(18.39,37.38)

25.6(16.92,34.36)

28.1(19.44,36.77)

23.4(14.75,31.95)

30.3(20.51,40.04)

21.7(13.07,30.25)

29.1(19.27,38.92)

22.7(14.43,31.05)

>750Empioyees(n-lll), %(95%Ci)

84.2(69.70,98.62)

68.1(53.13,83.14)

66.1(49.15,83.10)

42.1(23.80,60.45)

43.0(24.71,61.35)

54.3(35.18,73.39)

56.1(37.14,75.14)

81.5(71.80,91.17)

27.6(13.20,42.02)

16.8(6.97,26.72)

70.2(55.57,84,85)

70.2(54.99,85.46)

84,9(73.16,96.63)

80,5(68,00,93.01)

70.7(54,39,86,94)

48.2(28.63,67.73)

39.4(19.10,59.66)

28,3(14,62,41.88)

23,2(11,51,34,95)

29.6(13,94,45,26)

32.3(15.71,48.96)

50,9 (31,34,70.36)

29,3(9,53,49.06)

16.6(7.70,25.56)

41.4(21.23,61.49)

Note. Cl=confidence interval,

"Nonsignificant between-group difference.

August 2 0 0 8 , Vol 9 8 , No. 8 | American Journal of Public Health Linnan et al. \ Peer Reviewed | Research arid Practice | 1 5 0 5

RESEARCH AND PRACTICE

offered chronic obstmctive pulmonary dis- ease programs. Sites with more than 750 em- pioyees were more likely to offer cardiovascu- lar disease (50.9%), diabetes (48.2%), and higb-risk pregnancy (41.4%) programs than were other worksites. Smaller worksites were less likely to offer all types of disease manage- ment programs (Table 1). Differences accord- ing to worksite size were significant in the case of obesity, cardiovascular disease, high- risk pregnancy, and diabetes programs.

Work Environment Programs or Policies

With respect to providing an environment supporting physical activity, 27.6% of work- sites offered on-site shower facilities, 14.6% had an on-site fitness facility, 13.5% offered fitness or walking trails, and 6.2% provided signage to encourage stair use. Sites with larger numbers of employees were more likely to offer a supportive environment for physical activity. For example, 63.8% of employers with more than 750 employees

offered shower facilities (vs 20.9% of those with 50 to 99 employees), 49.6% (vs 9.8%) offered an on-site fitness facility, 40.5% (vs 7.7%) offered a fitness or walking trail, and 11.4% (vs 2.1%) promoted stair use with signage.

Overall, 2 4 % of worksites had a cafeteria (allowing them a chance to offer healthy food selections). Approximateiy 74% of sites with more than 750 employees had a cafeteria, as compared with 41.9% of sites with 250 to 749 employees, 24.5% of sites with 100 to 249 employees, and 12.9% of sites with 50 to 99 employees. Most sites (79.6%) had food or beverage vending services, with the largest sites more likely to provide such ser- vices. Overall, 37.4% of worksites reported labeling healthy food choices, and 5.6% of- fered promotions for healthy food choices (Table 2).

Approximately 4 0 % of worksites com- pletely prohibited smoking on worksite prop- erty, and 56.5% restricted smoking to outside

areas only; 12.4% provided employees fitness breaks at work. Overall, only 6.1% of sites of- fered catering policies to ensure that healthy food options were available at company events; 12.4% of sites with more than 750 employees reported having a catering policy in place. Overall, worksite policies prohibiting alcohol use (91.1%), drug use (93.4%), and firearm use (85.8%) were prevalent, whereas occupant protection policies for company ve- hicles (45%) were not. At each size category, the percentages of worksites that reported having a policy increased as the number of employees increased.

Programs and Policies by industry Type

In general, no differences in health promo- tion programs, activities, or screenings; disease management programs; work environments; or policies by industry type were observed. However, sites in the transportation/communi- cations/utilities and agriculture/mining/ construction categories were significantly less

TABLE 2-Seiected Viotk Environment and Poiicy Ciiaracteristics, by Worksite Size: Nationai Wori<site Heaitii Promotion Survey, 2004

Physical environment On-site fitness center On-site shower facilities Signage promoting stair use Fitness/walking trails Food/beverage services Cafeteria

Has a cafeteria Healthy food choices labeled

Special promotions offered

Policies

Fitness breaks provided

Catering policy

Smoking policy

Smoking completely prohibited

Smoking restricted to designated inside areas

Smoking restricted to outside areas

Alcohol use prohibited

Drug use prohibited

Occupant protection policy (vehicles)

Firearms prohibited

Incentives to promote participation

Total (n-730),

%(95%CI)

14.6(9.97,19.14)

27.6(22.87,32.36)

6.2(3.57,8.85)

13.5 (9.66,17.28)

79.6(74.5,84.7)

24.0(19.39,28.65)

37.4(26.32,48.56)

5.6(3.07,8.09)

12.4 (8.59,16.21)

6.1 (0.00,11.49)

39.9(34.12,45.65)

34.7(27.81,41.48)

56.5(49.24,63.77)

91.1(87.46,94.75)

93.4(90.30,96.54)

45.0 (39.18,50.98)

85.8(81.75,90.01)

25.9(20.0,31.82)

50-99 Employees,

%(95%CI)

9.8 (2.20,17.30)

20.9(13.59,28.15)

2.1(0.12,4.01)

7.7(2.17,13.13)

70.8(61.47,80.17)

12.9 (5.92,19.97)

34.6(6.50,62.75)

3.9(0.00,8.03)

11.0 (4.59,17.48)

6.3 (0.85,11.79)

34.2(24.66,43.73)

32.0(21.01,43.03)

50.8(38.90,62.67)

86.3(79.55,92.96)

91.8(86.37,97.36)

49.0(39.22,59.00)

83.0(75.66,90.43)

23.4(12.75,34.10)

100-249 Employees,

% (95% Cl)

13.17(5.63,20.71)

29.7(21.43,37.99)

11.7 (5.14,18.32)

13.9(7.22,20.64)

82.1(74.16,90.01)

24.5(17.02,31.95)

28.8(11.39,46.26)

5.4(1.37,9.42)

13.0(7.02,18.95)

5.7(1.37,9.93)

45.6 (36.52,54.61)

36.4(25.25,47.46)

56.5(44.59,68.35)

93.2(87.67,98.75)

94.4(89.46,99.30)

38.9(29.83,48.00)

87.5(81.09,93.97)

• 27.5(18.11,36.84)

250-749 Employees,

% (95% Cl)

17.5(9.46,25.50)

32.4(23.28,41.50)

4.2(1.57,6.74)

22.1(12.62,31.59)

95.9(92.67,99.21)

41.9(30.75,52.98)

32.4(16.50,48.37)

7.4 (3.62,11.12)

13.5(6.08,20.97)

4.7 (0.00,10.35)

40.8 (29.98,51.60)

39.3(25.88,52.73)

70.3(59.31,81.30)

98.5 (97.02,100.00)

94.2 (88.36,100.00)

45.6 (34.40,56.72)

87.4(79.72,95.05)

27.7(17.70,37.63)

> 750 Employees, %(95%CI)

49.6(29.98,69.24)

63.8(45.54,82.11)

11.4 (3.45,19.24)

40.5 (21.83,59.16)

95.4(91.12,99.61)

74.1(59.13,88.71)

73.1(53.64,92.63)

18.6(4.46,32.71)

17.6(4.21,31.37)

12.4(1.69,23.09)

48.5(28.91,68.26)

36.3(18.94,53.56)

77.4(64.17,90.61)

99.2(98.18,100.00)

99.2(98.18,100.00)

53.2 (35.54,71.03)

96.3(92.11,100.00)

28.7(12.17,46.22)

Note. Cl - confidence interval.

1506 I Research and Practice | Peer Revievi/ed | Linnan et ai. American Journal of Public Health | August 2008, Vol 98, No. 8

RESEARCH AND PRACTICE

TABLE 3-lncorporation of Key Elements of a Comprehensive Program, by Worksite Size: National Worksite Health Promotion Survey, 2004

Health education

Supportive social and physical

environment

integration

Linkage to reiated programs

Worksite screening

All 5 eiements

Total (n = 730),

% (95% CI)

26,2(21,54,30,84)

29,9(24,67,35,03)

28,6(23,37,33,74)

41,3(35,87,46,71)

23,5 (18,68,28,27)

6,9 (3,87,10,02)

50-99 Employees (n=179),

% (95% CI)

17,8(10,37,25,32)

24,0(15,28,32,73)

20,6(12,24,29,05)

29,6(20,68,38,43)

15,8 (8,07,23,49)

4,6(0,00,9,36)

100-249 Employees (n=229),'

% (95% CI)

26,2 (18,80,33,67)

32,5(24,40,40,68)

33,3(24,85,41,75)

43,7(34,66,52,70)

25,3(17,58,33,05)

6,0 (1,72,10,33)

250-749 Employees (n-211).

% (95% CI)

38,1(27,61,48,49)

33,5(23,43,43,63)

30,9(20,62,41,17)

59,3 (47,87,70,82)

30,5(20,99,39,96)

11,3 (3,80,18,76)

> 750 Employees (n-111),

% (95% Ci)

70,3(54,22,86,40)

53,7 (34,70,72,80)

61,4 (43,20,79,54)

80,5(65,61,95,36)

62,4(44,10,80,76)

24,1(4,03,44,21)

P

<,001

,04

,002

<,001

<,001

,03

Woie, CI = confidence inteivai,

likely to offer nutrition programs than were sites in the other industry categories, and sites in the agriculture/mining/construction cate- gory were less likely to offer diabetes screen- ing programs (data not shown).

Comprehensive Programs

Only 6,9% of worksites offered a compre- hensive worksite health promotion program (i,e,, a program incorporating all 5 key elements defined in Healthy People 2010; Tahle 3), Sig- nificcint differences according to worksite size were apparent with respect to provision of a comprehensive program; 24,1% of sites with more than 750 employees offered such a pro- gram, as compared with 11,3% of sites with 250 to 749 employees, 6,0% of sites with 100 to 249 employees, and 4,6% of sites with 50 to 99 employees. Sites in the manu- facturing (8,7%) and business/professional services (8,3%) categories were more likely to offer comprehensive programs than sites in the wholesale/retail (5,7%), transportation (2,9%), finance (2,4%), and agriculture/ mining (1,4%) categories.

When we examined each of the 5 key ele- ments individually, we found that linkages to related programs (41,3%) were most com- monly reported, followed by supportive social and physical environments (29,9%), integra- tion of the program into the organizational stiTjctiore (28,6%), health education (26,2%), and worksite screenings (23,5%), Worksites with fewer employees were less likely to offer a comprehensive program in general Eind were also less likely to offer any 1 of the 5 key ele- ments. Although 80,5% of worksites with more than 750 employees offered linkages to

TABLE 4-Relatlve Odds of Providing of a Comprehensive Health Promotion Program, by Worksite Characteristics: 2004 Nationai Worksite Health Promotion Survey

No, of empioyees

50-99 (Ref)

100-249

250-749

> 7 5 0

Experience'

Industry type

Manufacturing (Ref)

Finance

Whoiesaie/retaii

Transportation

Agricuiture/mining

Business/professionai services

Staff person in piace

Unadjusted OR (95% Ci)

1,00

1,34(0,35,5,14)

2,66 (0,70,10,13)

6,66 (1,42,31,23)*

0,59 (0,22,1,60)

1,00

0,26(0,09,0,73)*

0,63 (0,20,1,97)

0,31 (0,07,1,33)

0,15(0,03,0,86)*

0,94(0,31,2,83)'

29,86 (7,13,125,07)*

Muitivariate Adjusted OR (95% CI)

1,00

0,97(0,25,3,83)

1,75 (0,44,7,03)

4,41 (0,92,21,07)'

0,52 (0,21,1,35)

1,00

0,29(0,10,0,82)*

.1,06(0,31,3,61)

0,40 (0,09,1,90)

0,15(0,02,0,96)*

1,2(0,41,3,49)

10,26(1,97,53,41)*

Note. OR=odds ratio; CI=confidence interval,

"Number of years program had been in piace; the reference categoiy was programs in piace for iess than 5 years, *P<,05,

related programs, only 29,6% of those with 50 to 99 employees did so (P<,001),

Table 4 depicts the likelihood of providing a comprehensive worksite health promotion program according to worksite size, industry type, experience, and whether sites had a staff person dedicated to and responsible for health promotion. Both unadjusted (bivariate) and adjusted (multivariable) logistic regres- sion results are shown, along with point esti- mates and 9 5 % confidence intervals. Unad- justed results indicated that worksites with more than 750 employees were 6,7 times as likely as sites of all other sizes to offer a

comprehensive health promotion program and that sites in the agriculture and finance categories were significantly less likely than were sites in the other industry categories to offer a comprehensive program. Sites with a stiiff person dedicated to health promotion were nearly 30 times as likely to offer a com- prehensive program compared with sites without such a staff person.

The adjusted model showed that even after we controlled for worksite size, experi- ence, and industry type, sites with a dedi- cated staff person were 10,3 times more likely than were sites without a staff person

August 2008, Vol 98, No, 8 | American Journal of Public Health Linnan et al. \ Peer Reviewed | Research and Practice | 1507

RESEARCH AND PRACTICE

dedicated to health promotion to have a comprehensive worksite health promotion program (P=.O5; Table 4). In addition, the adjusted model showed that sites in the agri- culture and finance categories were signifi- cantly less likely than were sites in the other industry sectors to have a comprehensive program (P=.O5). Although worksites with more than 750 employees were 4.4 times as likely as sites of other sizes to have a com- prehensive program in place. Observed dif- ferences in worksite size only trended toward statistical significance (P=.O6).

DISCUSSION

One of the objectives of Healthy People 2010 was for at least 7 5 % of worksites to offer a comprehensive health promotion pro- gram, yet only 6.9% of our responding work- sites met this criterion. Sites with more than 750 employees offered more health promo- tion programs, services, and screening pro- grams; had more health-promoting policies in place; and were more likely to have health- supportive work environments than were worksites with fewer employees. This pattern was consistent with previous national work- site surveys. '̂* Given that small businesses (those with fewer than 500 employees) repre- sent 99.7% of all US employers and employ 50.1% of the private-sector workforce,''' it is apparent that important opportunities to im- prove the public's health are being missed.'^

In previous worksite surveys, induding the 1999 survey, metrics different from those de- scribed for the present survey were used to define types and levels of health promotion programming.''*'" In the 2004 National Worksite Health Promotion Survey, we be- lieved that it was important to monitor progress according to the Healthy People 2010 definition of a "comprehensive" health promo- tion program.

However, we also used broad, more- traditional means of documenting the extent to which worksite health promotion programs are available. Specifically, 9.7% of respon- dents indicated that they offered health edu- cation programming, a supportive work envi- ronment, and worksite screening programs, whereas 16.7% of respondents reported that they offered at least health education

programming and a supportive work environ- ment (data not shown). Thus, even when more-inclusive definitions of health promotion programs were applied, a low percentage of worksites reported offering these programs. Moreover, significant differences by worksite size persisted.

Few differences in health promotion pro- gramming, policies, and work environments by industry type were observed. Worksites in the manufacturing and business categories were more likely to offer comprehensive pro- grams, but small employers in each industry category were less likely to offer nearly all types of programs and services. Thus, differ- ent types of worksites may require different types of health promotion programs, policies, and practices, and a special emphasis on smaller worksites is needed if these worksites are to reach Healthy People 2010 objectives.

Despite relatively stable levels of health promotion programming among sites with more than 750 employees, there was a no- ticeable decline from 1999 levels in program- ming among sites with fewer than 750 em- ployees. This result may reflect a true decrease in programming, may represent measurement error (minor changes in the wording of questions may have generated dif- ferent responses), or may demonstrate that different cross-sectiond survey samples (even nationally representative samples) produce different and difficult-to-compare results.

One observation that supports a true drop in the number of health promotion programs offered is that worksites in the present survey reported significantly more perceived barriers (on identical questions) to offering health pro- motion programs than did worksites in the 1999 survey.* In our survey, 63.5% of work- sites reported that lack of employee interest was a barrier to offering heedth promotion programs, as compiired with 49.6% of work- sites in 1999 (P=.OO3). Lack of resources was cited as a barrier by 63.4% of employers in 2004 and 36.8% in 1999 (P=.O2); lack of partidpation by high-risk employees, lack of managernent support, conflicts with work demands, and lack of access to data were also cited at significantiy higher rates. These re- sults add credence to the possibility that a real decrease in programming occurred be- tween 1999 and 2004.

In contrast to the observed drop in pro- gramming between 1999 and 2004, re- ported policies and environmental supports remained stable during this period. Almost 30% of worksites reported that their sodal and physical environment at work was sup- portive of health.

Approximately 30% of worksites reported that their health promotion program had been in existence for at least 6 years. Our re- sults also indicated that sites with a staff per- son dedicated to health promotion were sig- nificantiy more likely to report having a comprehensive health promotion program. Although all of the factors just described (dedicated staff, polides, environmental sup- ports, and experience) are signs of perma- nence, the present findings reveal significant room for improvement

The most significant shortcoming, one that has persisted over time, is that worksites with small numbers of employees are less likely and (potentially) less able than large employ- ers to offer health promotion programs. More work must be done with smaU businesses to make a "business case" for health promotion, to develop new methods for reaching employ- ees, and to determine the employer and em- ployee incentives (e.g., tax credits, benefit dis- counts) that are most effective in supporting worker health.

Limitations Our study had several strengths as well as

limitations. An importsmt strength is that our data were derived from a nationally rep- resentative Scimple of worksites, allowing tentative comparisons between the present survey and the 1999 survey. However, be- cause of the variability of weights in 2 0 0 4 (resulting from the disproportionate nature of the sampling, whereby smaller sites were selected with considerably lower probabili- ties than larger sites), the precision of the confidence intervals for industry compar- isons was less than ideal. In addition, re- sponse rates in surveys of the general popu- lation are declining, and moderate rates of nonresponse may have a negative effect on point estimates and comparisons over time. Although adjustments for nonresponse have partiy addressed this issue, nonresponse bias may remain.

1508 I Research and Practice | Peer Reviewed | Linnan et al. American Journal of Public Heaith | August 2008, Vol 98, No. 8

RESEARCH AND PRACTICE

Our data represent, in the case of each worksite, the opinions of a single individual aligned with management; thus, caution in in- terpreting our results is warranted given evi- dence suggesting that employees' perceptions of access to and participation in worksite health promotion programs may vary consid- erably from employers' perceptions.'^ In addi- tion, although a standardized survey adminis- tration protocol was used, certain survey items that included définitions (e.g., "program integration" and "linkage") may have taken on different meanings for respondents across and within worksites. Because respondents answered questions with respect to their par- ticular worksite (as opposed to the company to which the worksite belonged), their re- sponses may not reflect the situations assod- ated with all health promotion programs sponsored by a given company. Finally, data on intervention quality and program effects on employee health were not gathered in the survey.

Conclusions There is a need for regular monitoring and

implementation of evidence-based worksite health promotion and health protection pro- grams. Employers can use information gath- ered from such programs for benchmarking purposes as they work toward achieving the objectives of Healthy People 2010. At a time when health care costs and work demands are rising, it is disturbing that few health pro- motion programs are available to employees. Our results can also be used to create part- nerships between employers, employees, health plans, policjmiakers, and health organi- zations with the goal of mobilizing workplaces to improve the public's health. Additional re- search that helps identiiy or develop effective worksite-based interventions, particularly for small businesses, is essential. •

About the Authors Laura Linnan and Mike Bowling are with the Department of Health Behavior and Health Education, School of Public Health, University of North Carolina, Chapel Hill. Jennifer Childress and Garry Lindsay are with Partnership for Pre- vention, Washington, DC. Carter Blakey and Penelope Royall are with the Office of Disease Prevention and Health Promotion, Rockville, Md. At the time of the study, Stephanie Pronk and Sharon Wieker were with Watson Wyatt Worldwide, Minneapolis, Minn.

Requests for reprints should be sent to Laura Linnan, ScD, CHES, School of Public Health, University of North Carolina, CB #7440, Chapel Hill, NC 27599-7440 (e-mail: [email protected]).

This article was accepted February 26, 2007.

Contrihutors L. Linnan developed the research questions and created first drafts of the article. M. Bowling conducted all of the analyses and drafted key parts of the Methods and Results sections. J. Childress, G. Lindsay, C. Blakey, S. Pronk, S. Wieker, and P. Royall developed the survey instruments and implementation procedures and pro- vided extensive feedhack on all drafts of the article.

Acknowledgments The Robert Wood Johnson Foundation provided initial funding to Partnership for Prevention in support of the 2 0 0 4 National Worksite Health Promotion Survey.

The survey was a joint effort of Partnership for Pre- vention and Watson Wyatt Worldwide, with the sup- port of the United States Department of Health and Human Services. A workgroup convened by Partner- ship for Prevention guided the development, imple- mentation, and analysis of the survey; the workgroup included experts from the Centers for Disease Control and Prevention, the National Center for Health Statis- tics, the Office of Disease Prevention and Health Pro- motion, and the University of North Carolina at Chapel Hill.

We thank Glorian Sorensen for helpful feedback on early versions of this artide.

Note. The views and opinions presented in this arti- cle are solely those of the authors, and the authors as- sume full responsibility for any errors or misrepresenta- tions. Statements do not necessarily represent the official position of the US Department of Health and Human Services or any other federal department or agency.

Human Participant Protection No protocol approval was needed for this study.

References 1. Stokols D, Pelletier K, Fielding J. The ecology of work and health: research and policy directions for the promotion of employee health. Health Educ Q. 1996; 23:137-158.

2. Dejoy DM, Southern DJ. An integrative perspec- tive on worksite health promotion./Mei/. 1993;35: 1221-1230.

3. Sorensen G, Barbeu E. Steps to a healthier US workforce: integrating occupational health and safety and worksite health promotion: state of the sdence. Available at: http://www.cdc.gov/niosh/worklife/steps/ 2004/whitepapers.html. Accessed December 8, 2006.

4. Healthy People 2010: With Understanding and Im- proving Health and Objectives for Improving Health. Washington, DC: US Dept of Health and Human Ser- vices; 2000.

5. Partnership for Prevention. Healthy workforce 2010: an essential health promotion sourcebook for employers, large and small. Availahle at: http:// www.preventorg/images/stories/Files/publications/ Healthy_Workforce_2010.pdf Accessed June 13, 2006.

6. Dun & BradstreeL Million dollar databases. Avail- able at: http://www.dnbmdd.com. Accessed June 13, 2006.

7. 1992 National Survey of Worksite Health Promo- tion Activities: Summary Report. Washington, DC: US Dept of Health and Human Services; 1993.

8. 1999 National Worksite Health Promotion Activi- ties: Summary Report. Washington, DC: US Dept of Health and Human Services; 2000.

9. Gwartney PA. TTie Telephone Interviewer's Hand- book: How to Conduct Standardized Conversations. San Francisco, Calif: Jossey-Bass Publishers; 2007.

10. SAS 9.1.3 Help and Documentation. Cary, NC: SAS Institiite Inc; 2004.

11. An AB. Performing logistic regression on survey data with the new SURVEYLOCISTIC procedure. Available at: http://www.lexjansen.com/pharmasug/ 2002/proceed/sas/sas05.pdf. Accessed March 15, 2006.

12. Kleinbaum DG, Klein M. Logistic Regression: A Self-Learning Text. 2nd ed. New York, NY: Springer- Verlag; 2002.

13. American Association for Public Opinion Re- search. Standard definitions and final dispositions of case codes and outcome rates for surveys. Available at: http://www.aapor.org/pdfs/standarddefs_4.pdf. Ac- cessed November 1, 2006.

14. US Small Business Administration. Advocacy small business statistics and research. Available at: http://appl.sba.gov/faqs/faqindex.cfm?areaID=24. Ac- cessed May 22, 2006. '

15. Wilson MG, DeJoy DM, Jorgensen CM, Crump CJ. Health promotion programs in small worksites: results of a national survey. AmJ Health Promot 1999;13: 3 5 8 - 3 6 5 .

16. Phillips K, Stokols D, McMahon S, Grzywacz J. Strategies for health promotion in small business. AmJ Health Promot 2004;19(suppl):l-7

17 Grosch JW, Alterman T, Petersen MR, Murphy LR. Worksite health promotion programs in the US: factors assodated with availability and partidpation. AmJ Health Promot 1998;13:36-45.

August 2008, Vol 98, No. 8 | American Journal of Public Health Linnan et ai. | Peer Reviewed | Research and Practice | 1509