Monday, November 16, 2009

Risk Assessment and Screening for Sexually Transmitted Infections, HIV, and Hepatitis Virus Among Long-Distance Truck Drivers in New Mexico, 2004-2006

Abstract (Summary)

We examined sexually transmitted infection (STI), HIV, and hepatitis virus prevalence and risk behaviors among truck drivers. We asked participants about their risk behaviors, and we screened them for STIs, HIV, and hepatitis infections. We used logistic regression to identify factors associated with outcomes. Of the 652 enrolled participants, 21% reported sex with sex workers or casual partners in the prior year. Driving solo (odds ratio [OR]=15.04; 95% confidence interval [CI]=1.92, 117.53; P=.01), history of injection drug use (IDU; OR=2.69; 95% CI=1.19, 6.12; P=.02), and history of an STI (OR=2.47; 95% CI=1.19, 5.09; P=.01) were independently associated with high-risk sexual behaviors. Fourteen percent of participants reported drug use in the previous year, and 11% reported having ever injected drugs. Participants tested positive as follows: 54 for HCV antibodies (8.5%), 66 for hepatitis B anticore (anti-HBc) antibodies (10.4%), 8 for chlamydia (1.3%), 1 for gonorrhea (0.2%), 1 for syphilis (0.2%), and 1 for HIV (0.2%). History of injecting drugs (OR=26.91; 95% CI=11.61, 62.39; P<.01) and history of anti-HBc antibodies (OR=7.89; 95% CI=3.16, 19.68; P<.01) were associated with HCV infection. Our results suggest a need for hepatitis C screening and STI risk-reduction interventions in this population.

Copyright American Public Health Association Nov 2009

[Headnote]
Objectives. We examined sexually transmitted infection (STI), HIV, and hepatitis virus prevalence and risk behaviors among truck drivers.
Methods. We asked participants about their risk behaviors, and we screened them for STIs, HIV, and hepatitis infections. We used logistic regression to identify factors associated with outcomes.
Results. Of the 652 enrolled participants, 21% reported sex with sex workers or casual partners in the prior year. Driving solo (odds ratio [OR]=15.04; 95% confidence interval [CI]=1.92, 117.53; P=.01), history of injection drug use (IDU; OR=2.69; 95% CI=1.19, 6.12; P=.02), and history of an STI (OR=2.47; 95% CI=1.19, 5.09; P=.01) were independently associated with high-risk sexual behaviors. Fourteen percent of participants reported drug use in the previous year, and 11% reported having ever injected drugs. Participants tested positive as follows: 54 for HCV antibodies (8.5%), 66 for hepatitis B anticore (anti-HBc) antibodies (10.4%), 8 for chlamydia (1.3%), 1 for gonorrhea (0.2%), 1 for syphilis (0.2%), and 1 for HIV (0.2%). History of injecting drugs (OR=26.91; 95% CI=11.61, 62.39; P<.01) and history of anti-HBc antibodies (OR=7.89; 95% CI=3.16, 19.68; P<.01) were associated with HCV infection.
Conclusions. Our results suggest a need for hepatitis C screening and STI risk-reduction interventions in this population. (Am J Public Health. 2009;99: 2063-2068. doi:10.2105/AJPH.2008.145383)

Studies in Africa, Southeast Asia, Eastern Europe, and South America have linked long-distance truck drivers and commercial sex workers with the dissemination of sexually transmitted infections (STIs), including human immunodeficiency virus (HIV) infection. 1-15 Evidence suggests that the spread of HIV throughout central Africa was facilitated by truck traffic along the Kinshasa-Mombasa highway. 1,2 High rates of STIs and HIV have been observed among long-distance truck drivers in India and Bangladesh, where truck drivers are implicated in the spread of STIs and HIV into rural areas and areas surrounding international border crossings.7-11 Results from studies in Eastern Europe suggest risky behavior and increased syphilis rates among truck drivers.12,13 Additionally, studies among truck drivers in Brazil show low levels of perceived risk of infection despite high rates of syphilis and high levels of risky behaviors, e.g., unprotected sex with multiple partners, including commercial sex workers, and high levels of drug use.14,15

Little is known about the roles that longdistance truck drivers and sex workers at truck stops might play in spreading STIs or HIV in the United States. An ecological study in North Carolina examining reported syphilis cases during an outbreak found that the counties along interstate highways had higher syphilis rates than other counties in the state.16 The authors theorized that truck drivers and sex workers might have played a role in this finding, but there were no data to support this. A 1995 ethnographic study in Florida examining STI risk behaviors of truck drivers found low levels of perceived STI or HIV risk but high levels of risky behaviors.17 However, no laboratory studies were conducted; therefore, there are no data estimating the prevalence of STIs among long-distance truck drivers in the United States.

To examine the environments in which STIs, HIV, and hepatitis virus are transmitted, and to assess the prevalence of STIs, HIV infection, and hepatitis virus infection and risk behaviors among truck drivers, we conducted a risk assessment and screening for STIs, HIV, and hepatitis among truck drivers traveling through New Mexico.

METHODS

From December 2004 through March 2006, we used mobile clinic vans to conduct this study at a large trucking terminal in Albuquerque, New Mexico, and at 10 truck stops on interstate highways elsewhere in the state. The 10 truck stops ranged in size from small (parking capacity approximately 50) to very large (more than 400 parking places). Nine truck stops had 24-hour restaurants, 4 had private security patrolling the lots, and 2 were located on Indian reservations and associated with casinos. Seven truck stops were on Interstate 40, running east-west through New Mexico; the rest were located in the southern part of the state on Interstate 25, Interstate 10, and at the intersection of interstates 285 and 360.

Truck driverswere recruited via citizens' band radio and leaflets distributed at trucking venues. Any long-distance truck driver aged 21 years or older who had a valid commercial driver's license, who traveled interstate, and who did not return home nightly was eligible for the study. Participants provided verbal informed consent and completed a face-to-face interview conducted by a study team member. STI and hepatitis screening was conducted after the interview. Interviews lasted 15 to 30 minutes, and each participant was reimbursed $35 cash.

The structured interview form collected demographics, driving history (e.g., years working as a driver), sexual behavior (e.g., condom use, STI history), and drug and alcohol use. All interviews were conducted anonymously, with no identifying or locating information collected, and all interview forms were coded with unique numbers.

STI, HIV, and Hepatitis Screening

Blood and urine samples were collected from study participants at the conclusion of the interview. We used a nucleic acid amplification test (Aptima Combo 2, Gen-Probe Inc, San Diego, CA) to test urine samples for chlamydia and gonorrhea. Blood serum samples were tested for syphilis,HIV, hepatitis B virusmarkers, andHCV markers. Syphilis antibodieswere assayed using a rapid plasma reagin (RPR) assay (Wampole Impact Syphilis RPR Card Test, Inverness Medical ProfessionalDiagnostics,Waltham,MA); positive RPR tests were confirmed using a Treponema pallidum particle agglutination (TPPA) assay (Serodia-TP-PA, Fujirebio Diagnostics, Malvern, PA). HIV antibodies were assayed using an HIV enzyme immunoassay (EIA) (Vironostika HIV- 1Microelisa System, bioMe'rieux, Marcy l'Etoile, France); positive EIA results were confirmed using an HIV-1Western blot assay (Genetic Systems HIV-1Western Blot, Bio-Rad Laboratories, Hercules, CA). Hepatitis B surface antigen (HBsAg) was assayed using the Genetic Systems HBsAg 3.0 test (Bio-Rad Laboratories), and total antihepatitis B core antibodies (anti-HBc) were assayed using the ETI-AB-COREK PLUS assay (DiaSorin, Saluggia, Italy). HCV antibodies were assayed using the ORTHO HCV version 3.0 ELISA test (Ortho Clinical Diagnostics, Rochester, NY). Positive test results for total HCV antibody were reported with a signal-to-cutoff ratio. A signal-to-cutoff ratio of at least 3.8 is predictive of a true positive test result more than 95% of the time.18

All laboratory testing was performed by the Scientific Laboratory Division of the New Mexico Department of Health. No specimens were tested for drugs or alcohol.

Data Analysis

No identifying information was collected; thus, it was possible that drivers volunteered more than once. To identify potential duplicates, data were examined to identify drivers reporting the same response for age, gender, race, ethnicity, marital status, number of years driving, and home state. No potential duplicates were identified in this manner.

Truck drivers were classified by type of driver: company drivers (union and nonunion), lease drivers, and owners/operators. Company drivers are employees of union and nonunion shops; all loads, driving schedules, and routes are arranged for them by the company. For union employees, the driving schedule is usually a set route. For nonunion employees, the driving schedules and routes are highly variable and often are modified while the driver is on the road. Lease drivers own their own truck and lease it to 1 or more companies. These drivers have some flexibility in their driving schedules and routes, and they usually have their loads arranged for them by the company to which they lease their truck. Owner/operator drivers are completely independent; these drivers own their own trucks, arrange their own loads, and determine their own driving schedule and routes.

We used EpiInfo version 6 (Centers for Disease Control and Prevention, Atlanta, GA) and Intercooled Stata version 9 (StataCorp LP, College Station, TX) to conduct data analyses and all logistic regressions (descriptive, univariate, stratified, and multivariate). Numeric variables were analyzed as continuous variables. Age and number of years driving were also examined as categorical variables: age was divided into10- year age groups, and years driving was divided into groups of less than1year,1to5 years,6to10 years, and more than 10 years. For multivariate logistic regression analyses examining risk factors independently associated with having sex with a sex worker, drug use, and positive laboratory test results, all variables found by univariate analyses to be significantly associated with the outcomeat P£.05were included in analyses.

RESULTS

This study was conducted 2 to 3 times per month from December 2004 through March 2006 at the trucking terminal or a truck stop. A total of 652 drivers enrolled. Demographic characteristics are shown in Table 1. Most drivers resided in 44 of the contiguous 48 US states, with a few living in Canada. Thirty-nine (6%) had been driving for less than 1 year, and half of the drivers had been driving for more than 10 years (mean=13 years; range=1-48 years). Drivers reported being away from home a mean of 288 nights per year (range=60- 365 nights). However, the time away from home was not evenly distributed throughout the year. Some drivers (<10%)>

A number of differences were found when driver characteristics were examined by gender. Female drivers were significantly more likely than male drivers to have attended or completed college (odds ratio [OR]=2.79; 95% confidence interval [CI]=1.58, 4.93; P<.01) and to always drive as part of a team (OR=5.76; 95% CI=10.55, 19.45; P<.01). Of drivers always driving as part of a team, female drivers were significantly more likely to drive with their spouse or steady partner than male drivers were (OR=8.99; 95% CI=3.43, 24.16; P<.01). Female drivers also reported significantly fewer years driving than male drivers (mean 7.4 years for women vs 14.0 years for men; P<.01) and a lower mean annual income than men ($50000 vs $61000).

Of the 652 drivers in this study, 5% were union employees, 71% were nonunion employees, 19% were lease drivers, and 5% were owner/operators. Union drivers reported almost always being home on weekends and spent significantly less time away from home than any of the other 3 types of drivers (mean=219 nights away from home per year vs 291, 296, and 278 nights away from home per year for nonunion drivers, lease drivers, and owner/operators, respectively; P<.01). Union drivers were also significantly more likely to always drive as part of a team (45% vs 20%, 14%, and 3% respectively), to have health insurance (100% vs 73%, 51%, and 57% respectively), and to have paid sick leave (100%vs 21%, 5%, and 0%; P<.01 for all analyses).

Regarding health status, 31% of drivers reported their current health as fair or poor, with obesity, poor diet, and lack of exercise being common concerns. Health insurance coverage was reported by 67% of drivers, but only 19% had paid sick leave. For unionized drivers, health insurance was a benefit provided with employment. Most nonunion company drivers could purchase insurance through the company for which they worked; however, many said the cost was prohibitive. The high cost of insurance was also the main reason given by lease drivers and owner/operators for not purchasing health insurance coverage. Even though two thirds of the drivers had health insurance, they reported great difficulty accessing care and locating providers. As a result, drivers reported continuing to drive when ill (unless extremely ill) and using overthe- counter medications to alleviate symptoms

Alcohol Use, Drug Use, and Sexual Risk Behaviors

Twenty-five percent of drivers reported no alcohol consumption in the previous year, and 33% reported rarely drinking alcohol. Only 270 (41.4%) reported drinking 1 or more drinks per week (range=1-60 drinks/week). Binge drinking ([double dagger]5 drinks at 1 sitting) in the previous year was reported by 47% of drivers. Among these, 21% reported binge drinking at least 10 times in the previous year (mean=10; range=1-360).

The mean number of reported lifetime sexual partners was 48 (median=15; range=1 to>1000), and 3% of male drivers reported ever having sex with a man. Previous STI treatment was reported by 132 drivers (gonorrhea 12%, chlamydia 5%, herpes 2%, syphilis 1%, and human papillomavirus 1%).

Having sex with a sex worker in the previous 5 years was reported by 74 male drivers (13% of male drivers) and ranged from 1 time to more than 100 times. Forty-eight male drivers also reported having sex with a sex worker in the previous year (range=1-30 times). Sex with a casual partner (e.g., pick-up at bars, Internet contacts, another driver, truck stop employees) was reported by10 female and126 male drivers (21% of all drivers) for the previous 5 years and by 7 female and 73 male drivers (12% of all drivers) for the previous year. Among drivers reporting sex with a sex worker, 46% stated they used condoms less than half of the time, and 32% reported never having used condoms. Similar levels of condom use was reported for sex with a casual partner. Half the drivers reported that they had not used a condom the last time they had sex with a sex worker or casual partner. There were no statistically significant differences in reported condom use for drivers who reported having a spouse or steady partner.

In multivariate logistic regression, only being a solo driver (OR=15.04; 95% CI=1.92, 117.53; P=.01), having a history of current or prior injection drug use (IDU; OR=2.69; 95% CI=1.19, 6.12; P=.02), and having a history of an STI (OR=2.47: 95% CI=1.19, 5.09; P=.01) remained independently associated with having sex with a sex worker in the previous year (Table 2). Having a current partner (OR=0.44; 95% CI=0.22, 0.89; P=.02) and increasing number of years driving (OR=0.95; 95% CI=0.92, 0.99; P=.01) remained independently associated with a decreased likelihood of having sex with a sex worker in the previous year.

Drivers were asked about their use of ecstasy, heroin, crack cocaine, powder cocaine, methamphetamines, and marijuana in the previous 1 and 5 years. Overall, 195 (30%) reported any drug use in the previous 5 years and 93 (14%) in the previous year. However, 25% of the 652 drivers had been driving less than 5 years and 6% for less than 1 year. Thus, analysis of reported drug use among working drivers included only those working during the entire time period being analyzed, i.e., 486 drivers working at least 5 years and 613 drivers working at least 1 year.

For the 486 drivers working at least 5 years, 126 (26%) reported any drug use during those 5 years, and 65 (14%) reported drug use in the previous 1 year. This significant decrease in reported drug use (P<.01 for marijuana and methamphetamine; P=.03 for powder cocaine) was reported to be a direct result of required random drug testing20 and the consequences of a positive test result, i.e., loss of job. For the 613 drivers working at least 1 year, 85 (14%) reported any drug use during that year (Table 3). Of drivers reporting any drug use in the previous year, 34 reported using marijuana at least monthly and 19 weekly, 17 used methamphetamines at least monthly and14 weekly, 7 used cocaine at least monthly, and 7 used crack at least monthly. A history of IDU was reported by 70 (11%) of all 652 drivers, and IDU in the previous year was reported by 9 (1%) drivers. Of the 9 drivers reporting recent IDU, 3 reported usingmultiple drugs, 4 injected heroin, 4 injected cocaine, and 4 injected methamphetamines.

Univariate and multivariate logistic regression analyses were used to identify risk factors associated with reported drug use in the previous year. Only consuming at least 1 drink per week in the previous year (OR=2.44; 95% CI=1.26, 4.71; P=.01) remained independently associated with increased likelihood of using drugs, and only having health insurance (OR=0.59; 95% CI=0.35, 0.99; P=.05) remained independently associated with a decreased risk of drug use (Table 2).

STI and Hepatitis Screening Test Results

Urine specimens were obtained from 631 (96.8%) drivers and were tested for chlamydia and gonorrhea. Blood specimens were obtained from 636 (97.5%) drivers and were tested for HIV, syphilis, and hepatitis B and C. One male driver had a positive test for gonorrhea, and 8 drivers (4 men, 4 women) had positive test results for chlamydia (Table 4). One man with a history of IDU had positive test results for HIV (ELISA and Western blot). One man with a history of prior treatment of syphilis had a reactive test for syphilis (RPR 1:4 and TPPA positive). Three male drivers were HBsAg positive, and 66 others (10.4%; 5 women and 61 men) had remote resolved hepatitis B virus infection, i.e., a positive serum anti- HBc antibody test with a negative HBsAg test.

A total of 54 drivers (8.5%; 4 women and 50 men) had a positive test result for HCV antibodies by EIA. The anti-HCV signal-tocutoff ratio for all but 1 of these drivers was greater than 4.2, indicating a high likelihood that the EIA results were true positive results.18 Of these 54 drivers, 36 (66.7%) reported prior or current IDU (2 of these 36 drivers also reported receiving blood transfusions prior to 1990, and 1 other had worked as an emergency medical technician); 2 reported blood transfusions prior to 1990; 1 worked as a dialysis nurse; 1 had a spouse with HCV; and for 14 there was no risk identified in the interview. Five already knew they had HCV, and 1 had undergone treatment for HCV. Of the 54 drivers positive for HCV, 28 (52%) reported having health insurance, 43 (80%) reported drinking alcohol in the previous year, 27 (50%) reported at least 1 episode of binge drinking in the previous year, and 11 (20%) reported binge drinking more than 10 times in the previous year. In multivariate analyses comparing drivers with positive HCV test results to those with negative results, history of IDU (OR=26.91; 95% CI=11.61, 62.39; P<.01) and having a positive anti-HBc antibody test (OR=7.89; 95% CI=3.16, 19.68; P<.01) were highly associated with positive HCV test results (Table 2).

DISCUSSION

Similar to results from studies of truck drivers in developing countries and in Florida, drivers in this study reported engaging in risky behaviors associated with STIs.1-15,17 However, STI prevalence was low. Multiple factors may contribute to this finding. First, rates of STIs in the United States among the general population are lower than in developing countries, particularly for countries with high rates of HIV infection. Second, sexual contact with commercial sex workers was reported by a small minority of study participants. Additionally, half of the study participants reported using a condom during their last sexual encounter. Finally, the sample was a voluntary, convenience sample; it is possible that those with STIs choose not to volunteer.

An interesting finding was the high prevalence of HCV infection. The rate in this sample, 8.5%, was higher than the highest rate (4.3% in people aged 40-49 years) reported in the third National Health and Nutrition Examination Survey (1988-1994).21Two thirds of the HCVantibody- positive truck drivers reported previous or current IDU as a risk behavior likely related to HCV infection. Most participants who were HCV positive were unaware of their infection. This finding is significant because drivers also reported ongoing alcohol consumption, including binge drinking. Alcohol consumption is a strong independent risk factor for the progression of HCV-associated liver disease that potentially can be modified through patient education and behavior change.22,23

We found that14%of drivers reported using an illicit drug during the previous year, with marijuana use reported more than twice as often as methamphetamine or cocaine use. Although no drug or alcohol testing was included in this study, the self-reported drug use among study participants was similar to the results of voluntary drug testing in Oregon in April 2007, when 10% of drivers tested positive for controlled substances.24 The Oregon study lent some validity to the self-reported drug use of drivers in the current study. Although most drivers in the current study reported occasional drug use, some reported regular use, including IDU. This presents a potentially grave safety issue if any drivers are under the influence while driving.

Access to health care was repeatedly cited as a major concern for the truck drivers, a finding that has been previously documented.25 Even though most drivers had health insurance, they reported barriers to use. Their jobs necessitate mobility and routines that impede scheduling medical or dental appointments. Efforts such as establishing networks of medical or dental facilities with flexible hours at locations near trucking venues may provide improved access to health services for this population.

This study has several limitations. The sample was a convenience sample of drivers traveling across New Mexico who volunteered to participate. Therefore, the results may not be representative of all US truck drivers. No information was available for drivers who did not volunteer. Behaviors of these drivers may be different from those who volunteered; drivers engaging in risky behaviors may have chosen not to participate. The demographic breakdown of drivers in this study was very similar to that estimated by the Department of Labor and the American Trucking Association. In 2000, the American Trucking Association estimated that 80% of truckers were White, 9.7% Hispanic, 11.7% African American, 25.7% did not complete high school, 59% were aged 35 to 54 years, and 4.6% were women.26 The Department of Labor estimates that up to 12% of drivers are women.27 However, there may be significant differences in risk factors or STI prevalence between drivers traveling through New Mexico and those traveling the East Coast or between cities with high rates of STIs or HIV.

Additionally, the study design included a face-to-face interview, and data were selfreported. Therefore, social desirability may have caused under- or overreporting of risky behaviors, and drivers may have been reluctant to report use of alcohol or illegal drugs.

Despite these limitations, this study contributes to the literature on STIs and HIV among US truck drivers by providing information on drivers' current health needs and opportunities for intervention among them. Prevalence of STIs and HIV was low, but drivers reported risky behaviors. Our results suggest that drivers may benefit from HIV, STI, and hepatitis prevention interventions embedded within comprehensive wellness programs that are convenient and easily integrated into the mobile environment of the trucking industry. Additional studies including different US trucker populations and more rigorous study designs should be conducted to confirm these results and provide more data to inform the development of STI and HIV intervention and wellness programs for the study population.

[Reference] » View reference page with links
References
1. Bwayo J, Plummer F, Omari M, et al. Human immunodeficiency virus infection in long-distance truck drivers in East Africa. Arch Intern Med. 1994;154:1391-1396.
2. Nzyuko S, Lurie P, McFarland W, Leyden W, Nyamwaya D, Mandel JS. Adolescent sexual behavior along the Trans-Africa Highway in Kenya. AIDS. 1997; 11(suppl 1):S21-S26.
3. Carswell JW, Lloyd G, Howells J. Prevalence of HIV- 1 in east African lorry drivers. AIDS. 1989;3:759-761.
4. Karim QS, Karim SS, Soldan K, Zondi Ml. Reducing the risk of HIV infection among South African sex workers: socioeconomic and gender barriers. Am J Public Health. 1995;85:1521-1525.
5. Ramjee G, Gouws E. Prevalence of HIV among truck drivers visiting sex workers in KwaZulu-Natal, South Africa. Sex Transm Dis. 2002;29:44-49.
6. Morris M, Podhisita C, Wawer MJ, Handcock MS. Bridge populations in the spread of HIV/AIDS in Thailand. AIDS. 1996;10:1265-1271.
7. Singh YN, Malaviya AN. Long distance truck drivers in India: HIV infection and their possible role in disseminating HIV into rural areas. Int J STD AIDS. 1994;5:137-138.
8. Manjunath JV, Thappa DM, Jaisankar TJ. Sexually transmitted diseases and sexual lifestyles of long-distance truck drivers: a clinico-epidemiologic study in south India. Int J STD AIDS. 2002;13:612-617.
9. Gibney L, Saquib N, Metzger J, Choudhury P, Siddiqui M, Hassan M. Human immunodeficiency virus, hepatitis B, C and D in Bangladesh's trucking industry: prevalence and risk factors. Int J Epidemiol. 2001;30:878-884.
10. Gibney L, Macaluso M, Kirk K, et al. Prevalence of infectious diseases in Bangladeshi women living adjacent to a truck stand: HIV/STD/hepatitis/genital tract infections. Sex Transm Infect. 2001;77:344-350.
11. Gibney L, Saquib N, Macaluso M, et al. STD in Bangladesh's trucking industry: prevalence and risk factors. Sex Transm Infect. 2002;78:31-36.
12. Grgic-Vitek M, Klavs I, Potocnik M, Rogl-Butina M. Syphilis epidemic in Slovenia influenced by syphilis epidemic in the Russian Federation and other newly independent states. Int J STD AIDS. 2002;13(suppl 1): 2-4.
13. Kulis M, Chawla M, Kozierkiewicz A, SubataTruck Drivers E, Sex Casual: An Inquiry Into the Potential Spread of HIV/AIDS in the Baltic Region. Washington, DC: World Bank; 2004.
14. Lacerda R, Gravato N, McFarland W, et al. Truck drivers in Brazil: prevalence of HIV and other sexually transmitted diseases, risk behavior and potential for spread of infection. AIDS. 1997;11(suppl 1):S15-S19.
15. Malta M, Bastos FI, Pereira-Koller EM, Cunja MD, Marques C, Strathdee SA. A qualitative assessment of long distance truck drivers' vulnerability to HIV/AIDS in Itajai, southern Brazil. AIDS Care. 2006;18:489-496.
16. Cook RL, Royce RA, Thomas JC, Hanusa BH. What's driving an epidemic? The spread of syphilis along an interstate highway in rural North Carolina. Am J Public Health. 1999;89:369-373.
17. Stratford D, Ellerbrock TV, Akens JK, Hall HL. Highway cowboys, old hands, and Christian truckers: risk behavior for human immunodeficiency virus infection among long haul truckers in Florida. Soc Sci Med. 2000;50:737-749.
18. Alter MJ, Kuhnert WL, Finelli L, Centers for Disease Control and Prevention. Guidelines for laboratory testing and result reporting of antibody to Hepatitis C virus. MMWR Recomm Rep. 2003;52(RR03):1-16.
19. Federal Motor Carrier Safety Administration. Hours-of-Service Regulations. Available at: http:// www.fmcsa.dot.gov/rules-regulations/topics/hos/hos- 2005.htm. Accessed June 15, 2008.
20. Federal Motor Carrier Safety Administration. Alcohol and Drug Rules. Available at: http://www.fmcsa. dot.gov/rules-regulations/topics/drug/engtesting.htm. Accessed June 15, 2008.
21. Armstrong GL, Wasley A, Simard EP, McQuillan GM, Kuhnert WL, Alter MJ. The prevalence of Hepatitis C virus infection in the United States, 1999 through 2002. Ann Intern Med. 2006;144:705-714.
22. Bell BP, Manos MM, Zaman A, et al. The epidemiology of newly diagnosed chronic liver disease in gastroenterology practices in the United States: results from population-based surveillance. Am J Gastroenterol. 2008;103:2727-2736.
23. Bedogni G, Miglioli L, Masutti F, et al. Natural course of chronic HCV and HBV infection and role of alcohol in the general population: the Dionysos Study. Am J Gastroenterol. 2008;103:2248-2253.
24. Oregon State Police. After Action Report: Operation Trucker Check-12, Woodburn Port of Entry, April 10- 12, 2007. Available at: www.oregon.gov/OSP/PATROL/ docs/trucker_check_after_action_report.pdf. Accesed June 15, 2008.
25. Solomon AJ, Doucette JT, Garland E, McGinn T. Health care and the long haul: long distance truck drivers-a medically underserved population. Am J Ind Med. 2004;46:463-471.
26. American Trucking Associations. The US Truck Driver Shortage: Analysis and Forecast. Available at: http://www.thetruckersreport.com/truckernews/ATADriverShortageStudy05[ 1].pdf. Accessed June 15, 2008.
27. Bureau of Labor Statistics, US Department of Labor. Occupational Employment Statistics. Available at: http:// www.bls.gov/cps/wlf-table14-2007.pdf. Accessed June 15, 2008.

[Author Affiliation]
Sarah Valway, DMD, MPH, Steven Jenison, MD, Nick Keller, BS, Jaime Vega-Hernandez, and Donna Hubbard McCree, PhD, MPH, RPh

[Author Affiliation]
About the Authors
Sarah Valway, Steven Jenison, Nick Keller, and Jaime Vega-Hernandez are with the Public Health Division, New Mexico Department of Health, Santa Fe. Donna Hubbard- McCree is with the National Center for HIV, Viral Hepatitis, STD, and TB Prevention, Centers for Disease Control and Prevention, Atlanta, GA.
Correspondence should be sent to Sarah Valway, DMD, MPH, New Mexico Department of Health, Infectious Diseases Bureau, 1190 South Saint Francis Drive, Santa Fe, NM 87502 (e-mail: s.valway@att.net). Reprints can be ordered at http://www.ajph.org by clicking the ''Reprints/ Eprints'' link.
This article was accepted February 17, 2009.
Contributors
All the authors conceptualized the study design and developed and tested questionnaires prior to implementation. N. Keller led the recruitment of truck drivers. S. Valway, S. Jenison, N. Keller, and J. Vega-Hernandez conducted interviews with study participants and oversaw specimen collection. S. Valway led data management, article writing, and analyses. S. Jenison, N. Keller, J. Vega-Hernandez, and D. Hubbard McCree made significant contributions to data management, article writing, and analyses. S. Jenison and D. Hubbard McCree provided overall supervision of the study.
Acknowledgments
This research was funded by cooperative agreement from the Centers for Disease Control and Prevention through the Association for Prevention Teaching and Research (U36/CCU300860).
The authors appreciate the assistance of the truck drivers who participated in this study. Without their assistance, the study could not have been completed.We also acknowledge the many staff of the New Mexico STD Program, whose assistance with this study was vital to its success.
Human Participant Protection
The study was approved by the institutional review boards of the Centers for Disease Control and Prevention and the New Mexico Department of Health.

References

Indexing (document details)

Subjects:Human immunodeficiency virus--HIV, Trucking industry, Truck drivers, Wellness programs, Studies, Risk factors, Hepatitis, Health services, Health care access, Flexible hours, Ethnicity, Drug testing, Data analysis, Alcohol use
Author(s):Sarah Valway, Steven Jenison, Nick Keller, Jaime Vega-Hernandez, Donna Hubbard McCree
Author Affiliation:Sarah Valway, DMD, MPH, Steven Jenison, MD, Nick Keller, BS, Jaime Vega-Hernandez, and Donna Hubbard McCree, PhD, MPH, RPh

About the Authors
Sarah Valway, Steven Jenison, Nick Keller, and Jaime Vega-Hernandez are with the Public Health Division, New Mexico Department of Health, Santa Fe. Donna Hubbard- McCree is with the National Center for HIV, Viral Hepatitis, STD, and TB Prevention, Centers for Disease Control and Prevention, Atlanta, GA.
Correspondence should be sent to Sarah Valway, DMD, MPH, New Mexico Department of Health, Infectious Diseases Bureau, 1190 South Saint Francis Drive, Santa Fe, NM 87502 (e-mail: s.valway@att.net). Reprints can be ordered at http://www.ajph.org by clicking the ''Reprints/ Eprints'' link.
This article was accepted February 17, 2009.
Contributors
All the authors conceptualized the study design and developed and tested questionnaires prior to implementation. N. Keller led the recruitment of truck drivers. S. Valway, S. Jenison, N. Keller, and J. Vega-Hernandez conducted interviews with study participants and oversaw specimen collection. S. Valway led data management, article writing, and analyses. S. Jenison, N. Keller, J. Vega-Hernandez, and D. Hubbard McCree made significant contributions to data management, article writing, and analyses. S. Jenison and D. Hubbard McCree provided overall supervision of the study.
Acknowledgments
This research was funded by cooperative agreement from the Centers for Disease Control and Prevention through the Association for Prevention Teaching and Research (U36/CCU300860).
The authors appreciate the assistance of the truck drivers who participated in this study. Without their assistance, the study could not have been completed.We also acknowledge the many staff of the New Mexico STD Program, whose assistance with this study was vital to its success.
Human Participant Protection
The study was approved by the institutional review boards of the Centers for Disease Control and Prevention and the New Mexico Department of Health.
Document types:Feature
Document features:Tables, References
Section:RESEARCH AND PRACTICE
Publication title:American Journal of Public Health. Washington: Nov 2009. Vol. 99, Iss. 11; pg. 2063, 6 pgs
Source type:Periodical
ISSN:00900036
ProQuest document ID:1897707891
Text Word Count4757
Document URL:http://proquest.umi.com/pqdweb?did=1897707891&sid=1&Fmt=3&clientId=45625&RQT=309&VName=PQD

Candidiasis Genetics; Researchers at University of Kansas

Candidiasis Genetics; Researchers at University of Kansas have published new data on candidiasi genetics

Abstract (Summary)

Drosophila cuticular hydrocarbons (CHCs) can function as pheromones and consequently affect mate recognition. In a previous study of the two major CHCs in females that affect mating discrimination between Drosophila simulans and D. sechellia, quantitative trait loci (QTL) were identified on the X and third chromosome, and a few candidate genes were potentially implicated. Here we specifically test candidate genes for CHC biosynthesis and determine the genetic architecture of four additional CHCs that differ in abundance between D. simulans and D. sechellia females. The same QTL, and new ones, were found for additional CHCs. By examining all these CHCs and exploring their covariance, we were able to ascribe putative function to the major QTL. Although desaturases have received considerable attention for their role in CHC biosynthesis, evidence here implies that elongases may be just as important.

(c)Copyright 2009, Health & Medicine Week via NewsRx.com

2009 NOV 9 - ( NewsRx.com) -- A new study, 'Identification of quantitative trait loci function through analysis of multiple cuticular hydrocarbons differing between Drosophila simulans and Drosophila sechellia females,' is now available (see also Candidiasis Genetics). "The genetics of sexual isolation, behavioral differences between species that prevent mating, is understood poorly. Pheromonal differences between species can influence sexual isolation in many animals and in some cases a single locus can cause large functional changes in pheromonal mating signals," investigators in the United States report.

"Drosophila cuticular hydrocarbons (CHCs) can function as pheromones and consequently affect mate recognition. In a previous study of the two major CHCs in females that affect mating discrimination between Drosophila simulans and D. sechellia, quantitative trait loci (QTL) were identified on the X and third chromosome, and a few candidate genes were potentially implicated. Here we specifically test candidate genes for CHC biosynthesis and determine the genetic architecture of four additional CHCs that differ in abundance between D. simulans and D. sechellia females. The same QTL, and new ones, were found for additional CHCs. By examining all these CHCs and exploring their covariance, we were able to ascribe putative function to the major QTL. Although desaturases have received considerable attention for their role in CHC biosynthesis, evidence here implies that elongases may be just as important. Sex determination genes do not seem to have a role in this species difference although D. sechellia is sexually dimorphic in CHCs, whereas D. simulans is not," wrote J.M. Gleason and colleagues, University of Kansas.

The researchers concluded: "Epistatic interactions, only detected for CHCs limited to D. sechellia, imply that complex interactions among loci may also be having a role in these compounds that affect mating isolation."

Gleason and colleagues published their study in Heredity (Identification of quantitative trait loci function through analysis of multiple cuticular hydrocarbons differing between Drosophila simulans and Drosophila sechellia females. Heredity, 2009;103(5):416-24).

For additional information, contact J.M. Gleason, University of Kansas, Dept. of Ecology and Evolutionary Biology, Lawrence, KS USA..

The publisher of the journal Heredity can be contacted at: Nature Publishing Group, 345 Park Avenue South, New York, NY 10010-1707, USA.

Keywords: United States, Lawrence, Candidiasis Genetics, Behavior, Vaginal Candidiasis.

This article was prepared by Health & Medicine Week editors from staff and other reports. Copyright 2009, Health & Medicine Week via NewsRx.com.

Indexing (document details)

Author(s):Anonymous
Document types:Expanded Reporting
Publication title:Health & Medicine Week. Atlanta: Nov 9, 2009. pg. 837
Source type:Periodical
ISSN:15316459
ProQuest document ID:1892898141
Text Word Count392
Document URL:http://proquest.umi.com/pqdweb?did=1892898141&sid=2&Fmt=3&clientId=45625&RQT=309&VName=PQD

Sunday, November 8, 2009

Data mining of tuberculosis patient data using multiple correspondence analysis

Data mining of tuberculosis patient data using multiple correspondence analysis
T W RENNIE, W ROBERTS. Epidemiology and Infection. Cambridge: Dec 2009. Vol. 137, Iss. 12; pg. 1699, 6 pgs
Copyright © Cambridge University Press 2009

(ProQuest: ... denotes non-US-ASCII text omitted.)

Tuberculosis

INTRODUCTION

There has been a rise in tuberculosis (TB) notifications in the UK since 1987 [1]. However, excluding TB in London, rates of TB in the UK are relatively low and stable. In the context of North East (NE) London, high rates of TB are observed in some primary care trust areas (PCTs) whilst in others rates are relatively low [2]. This demonstrates the complexity of TB epidemiology in the UK and London and is suggestive of a range of factors that give rise to high rates of TB in specific geographical areas.

As TB is a notifiable disease specialist TB healthcare professionals report demographic and clinical variables of patients who are notified to public health authorities. The Enhanced Tuberculosis Surveillance (ETS) system was introduced in 1999 to aid notification [3]. These collected data show the different demographic and clinical profiles of patients observed in NE London and may account for variations in TB rates. This is a valuable health information source, for example, in identifying commissioning priorities for different PCTs. This requires appropriate statistical support and effective communication to decision makers [4, 5]. However, analysis of large amounts of data with a large proportion of categorical/nominal data (e.g. gender, ethnicity, etc.) that can display multiple associations may prove to be difficult to interpret if bivariate comparisons are made. Factor analysis and principal components analysis (PCA) are inappropriate methods of analysis for these data which include a mix of continuous and categorical data. Multiple correspondence analysis (MCA) is an analytical method that allows analysis of multiple categorical variables [6]. The usefulness of this method lies in its reduction of large quantities of data and inclusion of any number of categorical variables although it does not provide a statistical assessment of association. We demonstrate the use of MCA as a tool for performing epidemiological 'mapping' of TB patient variables. This may prove to be useful in identifying commissioning priorities in NE London.

METHOD

MCA

Greenacre [7] describes correspondence analysis (CA) in its simplest form as a two-way cross-tabulation summarizing the distribution of frequencies to display a data 'map' in two-dimensional space. MCA is the multivariable extension of CA that allows explanation of relationships between two or more variables [8]. By including more than two variables in this type of analysis the complexity is increased; relationships between variables are described in terms of the variance of data. As this technique involves categorical variables the variance of the data, specific to each variable category, can be plotted in dimensional space; the variance for each category can be 'averaged' to one point in space (the centroid). This results in two graphical outputs - object plots which show the spread of category data variance, and variable plots which can display joint category plots. The latter is useful in that entered variables may then be described by the proximity of variable categories to other categories' points, their inertia (degree of variance), and whether they lie along particular dimensions in common with other category points. This technique differs from PCA in that it permits analysis of multiple categorical variables [9]. However, continuous data, such as age, may be categorized and entered for analysis. For a full description of the use of MCA see Greenacre [6].

Analytical strategy

Data from the ETS dataset for NE London between the years 2002 and 2007 were selected and entered for analysis; this included data for seven PCTs. Denotified TB cases, where initial TB diagnosis was later changed, were excluded ( n =441). Data were entered into a data-frame in SPSS (version 14.0; SPSS Inc., USA) for analyses. After categorizing continuous data (patient age), data were entered for MCA using the 'optimal scaling' option in SPSS. A two-dimensional graphical output plot of data displaying variable categories was selected (variable plot - joint category plot function in SPSS).

RESULTS

Data for 4947 TB patients between the years 2002 and 2007 were entered for analysis. In this cohort of patients, male gender was slightly more common and the three most common ethnicities were Black African, Indian Asian, and Pakistani Asian; only 18·3% of patients were born in the UK (Table 1). A minority of patients (11·7%) had their consumption of treatment supervised by directly observed treatment (DOT) and over a third of patients were hospitalized. For three variables in particular (Table 1: employment, sputum smear test, bacterial resistance) there were large amounts of missing data. However, for data available, 46·4% of patients tested had a positive sputum smear result ( n =2269) and 18·3% of patients exhibited TB strains of any bacterial resistance to first-line TB medicines ( n =1673).

Table 1.

Demographic and clinical TB patient data, 2002-2006

DOT, Directly observed treatment.

Percentages calculated from n =4947 unless other sample size quoted due to missing data.

* Employment excluding children, retired, housewives, asylum seekers or any ambiguity regarding current employment.

[dagger]

'Any resistance' refers to resistance to any of isoniazid, rifampicin, streptomycin, ethambutol, pyrazinamide.

MCA was used to analyse these data in three ways: data were entered for analysis in their entirety, data were analysed by PCT, and data were analysed by year.

Complete dataset analysis

When all of the data were analysed together the joint category plot was complex and difficult to interpret reliably (Fig. 1 a ). However, PCT6 associated with 'Bangladeshi' ethnicity as an outlying group. This finding demonstrated the known higher prevalence of Bangladeshi TB patients in this PCT [10]. However, this strong association dominated the output. Therefore, to investigate associations between other variables without the dominating effect of ethnicity on PCT6, ethnicity was excluded and the analysis repeated (Fig. 1 b ). This suggested that both PCT6 and DOT ('Yes') categories were outliers from the dataset.

Fig. 1.

Multiple correspondence analysis graphical output of TB variable categories. ( a ) All variables, all years. ( b ) All variables, all years except ethnicity.

Analysis by PCT

MCA was repeated by analysing by each separate PCT. Figure 2 a displays an example of the output for PCT2 and is suggestive of an association between two groups of variable categories: Group 1: DOT ('Yes'), previous diagnosis ('Yes'), UK born ('Yes') and age >75 years. Group 2: hospital admission ('Yes'), positive sputum smear result ('Pos'), drug resistance ('Res') and pulmonary TB ('Yes'). The output for PCT7 is suggestive of a division between recent and earlier years of notification (Fig. 2 b ). More recent years (2004-2007) appear to group with more positive variable categories such as patients not being admitted to hospital ('No'), no previous diagnosis ('No') and no DOT ('No'). Earlier years (2002-2003) appear to associate with less positive variable categories such as previous TB diagnosis ('Yes'), DOT ('Yes'), and positive sputum smear result ('Pos').

Fig. 2.

Multiple correspondence analysis graphical output of TB variable categories by primary care trust (PCT). ( a ) PCT2; ( b ) PCT7.

Analysis by year

Finally, MCA was repeated by analysing by year. For example, Figure 3 a displays data from 2002 with a possible association between PCT2 and PCT3 with the variable categories DOT ('Yes') and UK born ('Yes'). However, in 2007 this specific grouping was not observed although PCT2 appeared to associate with DOT ('Yes'), previous TB diagnosis ('Yes'), UK born ('Yes') and resistance ('Res') suggesting a complex case-load for this PCT (Fig. 3 b ).

Fig. 3.

Multiple correspondence analysis graphical output of TB variable categories by year. ( a ) 2002; ( b ) 2007.

DISCUSSION

A commissioning framework report published by the Department of Health and informed by the governmental White Paper 'Our health, our care, our say: a new direction for community services' [4] highlighted the need for understanding the requirements of both populations and individuals as well as more effective sharing and use of information [5]. Data reported by healthcare systems is used to inform decisions concerning the commissioning of health services. Although these processes tend to be quite blunt, nevertheless, health commissioning would be ill-informed without the use of such data sources. It is pertinent to identify local trends in data to best focus healthcare resources and commission services appropriately. MCA is a well reported technique for the reduction of data and has previously been utilized in a wide range of different disciplines, e.g. analysis of wealth indices [9], more informative analyses of data for cardiac implantable devices [11], and investigations into subjective well-being, poverty and ethnicity [12]. This technique has previously been advocated for its use in the analysis of large datasets of categorical data, identifying themes according to data variance, and for scaling methods.

The current study used MCA to epidemiologically 'map' data that related to TB patients in NE London between 2002 and 2007. This identified a number of trends between data variables, differences between PCTs, and changes over time. For example, there appeared to be an association between patients that were born in the UK, patients that received DOT, and patients that were admitted to hospital. There may also be links between these variable categories and resistance to anti-tuberculous drugs and higher age group (>75 years). These associations are rational in that, within the TB population in London where most patients were born outside the UK, UK-born patients who contract TB are more likely to be older due to reactivation of disease rather than primary infection, and older patients are more likely to be admitted to hospital. MCA output for one PCT (PCT7) appeared to suggest recent improvement in that more positive variable categories, such as no hospital admission, associated closer to recent year categories (2004-2007) whereas less positive categories, such as previous TB diagnosis, associated closer to earlier year categories (2002-2003).

When analysed by year a similar grouping of more negative variable categories with two PCTs in particular (PCT2 and PCT3) was observed in 2002, and a similar grouping again observed for one of these PCTs (PCT2) in 2007. This suggests that cohorts of patients located within these PCTs had a greater burden of patients with complex needs in terms of provision of DOT and managing drug resistance, and that this issue had probably been resolved over time for PCT3. This clearly has resource implications. Treating patients with drug resistance, for example, has been estimated to be ten times the cost of treating a patient with drug-sensitive forms of TB [13]. Therefore, year-by-year analyses of this kind may inform where priorities lie. The various associations can be validated with further investigation to identify whether there are indeed greater priorities for certain PCTs in relation to specific patient groups and this, in turn, can inform commissioning priorities.

With such a large dataset where small associations are more likely to achieve statistical significance, MCA provides meaningful analyses that account for interactions between variables in the dataset as a whole. Another benefit of MCA is that it allows analysis of numerous variables of a categorical nature - the only continuous variable in the current study was patient age which was categorized for analysis. Analysis of a wider set of more descriptive variables in the current study, focusing on other aspects of patient complexity, for example, would better inform TB priorities for each PCT. However, to our knowledge, this is the first instance of an analysis of this type being performed with the explicit aim of identifying commissioning priorities. In addition, we believe this to be the first reporting of a TB dataset in this way.

A number of variables had large amounts of missing data. For example, sputum smear results and results for drug sensitivities were available for less than half of the cohort. This may relate to such results only having been recorded by TB services when they were deemed of clinical importance, such as resistance to a particular drug (recording only where tests had been performed and results obtained). However, it implies that these variables, in particular, were not reliably reported. Better recording of data would help to ensure that analyses were more reliable. Interestingly, MCA is a method used to explore patterns of missing data by categorizing missing data and including it in analyses, e.g. see Greenacre [7]. This technique could have been applied for the current dataset to assess whether missing data for specific variables differed from data that were better reported. Although this was beyond the aim of our study we are currently assessing data from the ETS dataset to better understand what the missing data might represent and, therefore, clarify reasons for non-reporting of data. In the current study only two-dimensional analyses were carried out to simplify the interpretation of results. In reality, the association between variables may be multidimensional and reveal further relationships between variable categories. However, for the purposes of using MCA as a commissioning tool multidimensional analyses are unlikely to be of significant added benefit.

In conclusion, we present an analytical technique that allows analysis of multiple datasets that can contain different data types. This tool can be used as an epidemiological method to inform commissioning priorities in healthcare such as TB service provision. Whilst users should be aware of the limitations, MCA is an efficient technique that effectively produces a data map displaying association. This may be of particular use where large amounts of heterogenous data are available.

ACKNOWLEDGMENTS

Our thanks to the continued efforts of TB Services in North East London without whose help this work could not have been conducted.

DECLARATION OF INTEREST

T.W.R. and W.R. are both employed on a full-time basis by the National Health Service.

[Reference]
REFERENCES
1. Health Protection Agency. (www.hpa.org.uk). Accessed 19 January 2009.
2. North East London TB Network. Annual report of demographic and epidemiological trends of TB in North East London. London, 2007.
3. BP Van. Data mining of tuberculosis patient data using multiple correspondence analysis Communicable Disease and Public Health 1998; 1: 219-220.
4. Department of Health. Our health, our care, our say: a new direction for community services. London: Department of Health, 2006.
5. Department of Health. Commissioning framework for health and well-being. London: Department of Health, 2007.
6. M Greenacre. Correspondence Analysis in Practice, 2nd edn. London: Taylor and Francis, 2007.
7. M Greenacre. Data mining of tuberculosis patient data using multiple correspondence analysis. Gaceta Sanitaria 2002; 16: 160-170.
8. E Kaciak, J Louviere. Data mining of tuberculosis patient data using multiple correspondence analysis. Journal of Marketing Research 1990; 27: 455-465.
9. LD Howe, JR Hargreaves, SR Huttly. Data mining of tuberculosis patient data using multiple correspondence analysis. Emerging Themes in Epidemiology 2008; 5: 3.
10. Directorate of Public Health. Tower Hamlets Public Health Report. Tower Hamlets Primary Care Trust, 2007.
11. M Guéguin, Data mining of tuberculosis patient data using multiple correspondence analysis. Conference Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society 2007; 1: 3848-3851.
12. DF Neff. Data mining of tuberculosis patient data using multiple correspondence analysis. Social Indicators Research 2007; 80: 313-341.
13. VL White, J Moore-Gillon. Data mining of tuberculosis patient data using multiple correspondence analysis. Thorax 2000; 55: 962-963.

[Author Affiliation]
North East London Tuberculosis Commissioning Unit, Newham Primary Care Trust, London, UK

Indexing (document details)

Author(s):T W RENNIE, W ROBERTS
Author Affiliation:North East London Tuberculosis Commissioning Unit, Newham Primary Care Trust, London, UK
Document types:Feature
Publication title:Epidemiology and Infection. Cambridge: Dec 2009. Vol. 137, Iss. 12; pg. 1699, 6 pgs
Source type:Periodical
ISSN:09502688
ProQuest document ID:1894220731
Text Word Count2479
DOI:10.1017/S0950268809002787
Document URL:http://proquest.umi.com/pqdweb?did=1894220731&sid=2&Fmt=3&clientId=45625&RQT=309&VName=PQD

Monday, January 26, 2009

Influence of Life Course Socioeconomic Position on Older Women's Health Behaviors: Findings From the British Women's Heart and Health Study

Influence of Life Course Socioeconomic Position on Older Women's Health Behaviors: Findings From the British Women's Heart and Health Study
Abstract (Summary)

We examined the association between health behaviors and socioeconomic status (SES) in childhood and adult life. Self-reported diet, smoking, and physical activity were determined among 3523 women aged 60 to 79 years recruited from general practices in 23 British towns from 1999 through 2001. The most affluent women reported eating more fruit, vegetables, chicken, and fish and less red or processed meat than did less affluent women. Affluent women were less likely to smoke and more likely to exercise. Life course SES did not influence the types of fat, bread, and milk consumed. Adult SES predicted consumption of all foods considered and predicted smoking and physical activity habits independently of childhood SES. Childhood SES predicted fruit and vegetable consumption independently of adult SES and, to a lesser extent, predicted physical activity. Downward social mobility over the life course was associated with poorer diets and reduced physical activity. Among older women, healthful eating and physical activity were associated with both current and childhood SES. Interventions designed to improve social inequalities in health behaviors should be applied during both childhood and adult life.

[Headnote]
Objectives. We examined the association between health behaviors and socioeconomic status (SES) in childhood and adult life.
Methods. Self-reported diet, smoking, and physical activity were determined among 3523 women aged 60 to 79 years recruited from general practices in 23 British towns from 1999 through 2001.
Results. The most affluent women reported eating more fruit, vegetables, chicken, and fish and less red or processed meat than did less affluent women. Affluent women were less likely to smoke and more likely to exercise. Life course SES did not influence the types of fat, bread, and milk consumed. Adult SES predicted consumption of all foods considered and predicted smoking and physical activity habits independently of childhood SES. Childhood SES predicted fruit and vegetable consumption independently of adult SES and, to a lesser extent, predicted physical activity. Downward social mobility over the life course was associated with poorer diets and reduced physical activity.
Conclusions. Among older women, healthful eating and physical activity were associated with both current and childhood SES. Interventions designed to improve social inequalities in health behaviors should be applied during both childhood and adult life. (Am J Public Health. 2009;99:320-327. doi:10.2105/AJPH.2007.129288)


In 1977, the United Kingdom Department of Health commissioned an inquiry focusing on health inequalities in the country's population. The resulting report-the Black Report, published in 1980-highlighted the marked association between adult socioeconomic status (SES) and mortality rates.1 Such socioeconomic gradients inmortality rates persist today, tracking into old age.2

Inequalities in health are a result of clearly identifiable social and economic factors that could potentially be modified to improve people's quality and length of life. Employment, education, housing, transportation, environment, health care, and "lifestyle" (in particular smoking, exercise, and diet) all affect health and tend to be favorably distributed in advantaged groups.

In the United Kingdom, the introduction of the National Service Framework for Coronary Heart Disease in 2000 was intended to reduce the prevalence of and social inequalities in coronary risk factors in the country's population. 3 Achieving these aims requires equitable access to and use of preventive care irrespective of SES, age, and gender. Health promotion initiatives such the "5-a-day" fruit and vegetable diet plan,4 smoking cessation clinics, and structured exercise plans have all been part of the drive to reduce the prevalence of coronary risk factors.

Recent years have seen increased recognition of the potential implications of life course SES and a deeper understanding of the conceptual framework on which it is based.5,6 There is growing evidence that coronary heart disease (CHD) risk is associated with life course SES,7-10 with those in the most disadvantaged SES groups throughout life showing nearly 3 times greater risk than those in more advantaged groups.8 This raises the question of the extent to which behavioral CHD risk factors are similarly dependent on life course SES. We examined the effects of childhood and adulthood SES on various health behaviors (diet, smoking, and physical activity) of older British women.

METHODS

Study Design and Data Collection

We conducted a cross-sectional analysis of baseline data from participants in the British Women's Heart and Health Study. The methodology of that study has been fully reported elsewhere.11 Briefly, from 1999 through 2001, 4286 women aged 60 to 79 years were recruited from general practice lists in 23 representative British towns. Participants completed a questionnaire including items focusing on diet, smoking,12 and physical activity. As a means of gathering dietary data, women were asked how often (more than once a day, daily, most days, once or twice a week, less than weekly, or never) they ate fresh fruit, green vegetables, meat, and other foods.

Behavioral Data

Principal-component analyses were used to identify various food groups. Fruits, salads, green vegetables, fish, and poultry formed the first component; however, given the public health focus on fruits and vegetables, these foods were examined separately. Red and processed meat formed the second component; healthful bread (e.g., whole-meal bread), milk (e.g., skim milk), and fat (e.g., vegetable oil rather than animal fats, and low-fat margarine rather than butter) formed the third.

Participants were asked to indicate the number of hours each week during the winter and summer they engaged in a specified range of physical activities; they were also asked to rate their walking speeds.13 These measurements were used to calculate their weekly number of hours of moderate or vigorous physical activity. Activities considered moderate or vigorous included walking at a relatively brisk or fast pace, cycling, heavy gardening, and other physical exercise (e.g., aerobics, swimming).

Socioeconomic Status Data

Ten SES indicators8 were used to construct a life course SES score and childhood and adult subscores: longest-held occupation of the participant's father during her childhood; whether the participant's childhood home had a bathroom and a hot water supply; whether the participant had shared a bedroom as a child; whether, during the participant's childhood, her family had access to a car; the age at which the participant completed full-time education; the longest-held occupation of the participant and her spouse; the participant's current housing status (whether she lived in rented social housing or owner-occupied and private rented properties); and the participant's current automobile access and pension arrangements (state only or state in combination with other arrangements).

Participants' childhood social class was based on their fathers' longest-held occupation, and their adult social class was based on their husbands' longest-held occupation (or, in the case of single women, their own longest-held occupation). Adult and childhood social class categories, defined according to the UK registrar general's classification, ranged from I (nonmanual, professional occupations) to V (manual, unskilled occupations). Given that the life-course SES score denoted the number of socioeconomic hardships experienced by women, a score of 10 indicated the greatest level of hardship.

We assessed the effects of changes in SES over the life course on health behaviors by classifying upward social mobility as change from manual social class in childhood (defined according to father's occupation) to nonmanual social class in adulthood (defined according to husband's occupation or, among unmarried women, their own occupation); downward social mobility was classified as the reverse circumstance. All analyses were restricted to women without any evidence of CHD or stroke at baseline (n=3523; 83% of the cohort); 595 women (13%) were excluded from the analyses (with the exception of those involving multiple imputations) because they also had missing data on 1 or more SES indicator.

Statistical Analyses

Women were grouped according to SES score, and the percentage of women reporting adverse health behaviors in each SES group was calculated. We assessed differences in the percentages of women reporting high-risk behaviors by individual SES indicator variables. In addition, stratifying by town of recruitment, we conducted logistic regression analyses examining the relative importance of childhood SES and adult SES scores as predictors of health behaviors and the effects of upward and downward social mobility on health behaviors.

We used conditional logistic regression in most of our analyses because the sampling strategy produced data clustered according to town of residence. We used ordinal logistic regression, clustered by town, in analyses of smoking and diet because the behavioral data were collected in 3 ordered categories. This technique allowed us to avoid using multiple significance tests, which would have been required to compare each pair of categories in turn.

Our analyses were based on the assumption that missing data were missing completely at random; that is, women included in the analyses could be regarded as a random sample of the women who took part in the study. If this assumption proved to be false, our results could be biased.

We assessed the sensitivity of the results by examining the effects of missing data. We assumed that data were missing at random (rather than missing completely at random, where the probability of data being missing does not depend on observed or unobserved values) and in this situation the missing values depend on the values of variables measured in the study. The missing values can then be imparted from knowledge of other measured values. Multiple imputation allowed our analysis to be conducted under the missing-atrandom assumption; we used the multivariate chained equation method,14,15 including all of the health habit variables and childhood and adulthood SES scores in the imputation model.

Ten regression switching cycles were used with 20 imputed data sets. Use of Rubin's formulas for combining results from the separate imputed data sets ensured that any incomplete data were properly accommodated in the inferences. The results of these alternative analyses were very similar to the results of the analyses conducted with women who had complete data (i.e., the analyses described here). Stata version 9 (StataCorp, College Station, TX) was used in conducting all analyses.

RESULTS

Table 1 presents health behavior data by SES score, and Table 2 shows differences in the percentages of women reporting unhealthful behaviors according to individual SES indicator variables. Table 3 shows odds of unhealthful behaviors for each 1-unit increase in childhood and adulthood SES score (i.e., increasing deprivation), with adjustment of childhood associations for adult SES (and vice versa).

Diet

Only 10% of women reported eating 4 or 5 portions of fruits and vegetables daily (the UK government recommendation); half reported consuming less than 2 portions (Table 1). A majority of the women (55%) selected mostly healthful fat, milk, and bread options; 30% ate red or processed meat on most days.

Women in the most deprived groups (those with an SES score of 9 or 10) had poorer diets than did women in the less deprived groups, consuming fewer fruits and vegetables (61% consumed less than 2 portions per day) and more red or processed meats (41% ate these meats on most days; Table 1). Both childhood and adult indicators of low SES were associated with unhealthful diets (Table 2). Eleven percent (P<.001) more women raised in manual social class families than in nonmanual social class families ate fruits and vegetables less than twice a day. Similar differences of between 5% and11%in consumption of fruit and vegetables were seen for other childhood deprivation indicators (e.g., no hot water in the family home, no family access to a car).

Adult indicators of deprivation showed similar levels of strength; the strongest predictor was current residence in local authority (i.e., social) housing (13% more women living in local authority housing than women not living in such housing reported eating fruit and vegetables less than twice a day; P<.001). Both childhood (for each 1-unit increase in childhood SES score, odds ratio [OR]=1.13; 95% confidence interval [CI]=1.07, 1.19) and adult (for each 1-unit increase in adult SES score, OR=1.16; 95% CI=1.07, 1.25) SES scores were independent predictors of fruit and vegetable intake (Table 3).

Associations with the other dietary variables were less strong. Consumption of red or processed meat on most days was independently associated with adult SES (adjusted OR=1.15; 95% CI=1.06, 1.25) but not childhood SES (adjusted OR=1.03; 95% CI=0.97, 1.09) after mutual adjustment. Poultry and fish consumption showed a similar pattern of stronger association with adult SES (adjusted OR=1.17; 95% CI=1.08, 1.26) than childhood SES (adjusted OR=1.06; 95% CI=1.00, 1.12). Selection of mostly healthful fat, milk, and bread options was not significantly related to SES or any SES subcomponents.

Smoking

Most women (57%) had never smoked tobacco regularly, and only 11% currently smoked (Table 1). However, 62% of women in the lowest SES group had smoked regularly at some point in their life (and 18% of them continued to smoke), as compared with less than 50% of the women in all other SES groups. The median age of smoking initiation was 18 years (5th percentile=15 years, 95th percentile=35 years), and there were minimal differences according to SES. Among quitters, women in higher SES groups quit at a younger median age (45 years [5th percentile=24 years, 95th percentile=65 years] among women with SES scores of 0-3 and 51 years [5th percentile=22 years, 95th percentile=70 years] among women with SES scores of 7-10).

Smoking was associated with having grown up in a manual social class family but was not related to other childhood SES indicators (Table 2). All adverse adult SES indicators were associated with smoking. The strongest predictor was local authority housing tenancy; women living in such housing were 18% more likely to smoke. As can be seen in Table 3, adult SES, but not childhood SES, was independently associated with smoking (for each 1-unit increase in adult SES, adjusted OR=1.18; 95% CI=1.09, 1.27).

Physical Activity

Most women were inactive. Sixty-one percent reported less than 2 hours per week of moderate or vigorous exercise; however, more than one quarter (28%) engaged in more than the recommended minimum of 3 hours per week. Generally, more women in the most disadvantaged SES groups than in the less disadvantaged groups reported a sedentary lifestyle (P<.001).

Adverse individual childhood and adult SES indicators were each associated with an increase in physical inactivity of at least 5% (Table 2). The strongest association was with local authority housing tenancy in adulthood; 16% more women living in this type of housing than in other types of housing engaged in less than 2 hours of moderate or vigorous activity each day. Adult SES and childhood SES were both independently associated with physical activity, but the association with adult SES was stronger (for each 1-unit increase in childhood SES, adjusted OR=1.06; 95% CI=1.01, 1.12; for each 1-unit increase in adult SES, adjusted OR=1.22; 95% CI=1.13, 1.32).

Social Mobility

Table 4 shows the effects of social mobility, classified according to father's and husband's social class (or, in the case of unmarried women, their own social class), on women's health behaviors. Upwardly mobile women were less likely to report unhealthful behaviors than were women who remained in the manual group. For example, they were 37% less likely to consume small amounts of fruits and vegetables (OR= 0.73; 95% CI=0.61, 0.88) and 21% less likely to be inactive (OR=0.79; 95% CI=0.66, 0.94).

Downwardly mobile women adopted worse health behaviors than women remaining in the nonmanual social class. For example, they were 51% more likely to eat red or processed meat on most days (OR=1.51; 95% CI=1.04, 2.18) and 47% more likely to engage in less than 2 hours of exercise per week (OR=1.47; 95% CI=1.05, 2.06); surprisingly, however, they were 45% more likely to select mostly healthful fat, milk, and bread options (OR=1.45; 95% CI=1.03, 2.06). We found no effects of social mobility on smoking. Although downward social mobility adversely affected women's diet and physical activity behaviors, the effect was not as marked as that observed when women who had remained in the manual social class throughout their life were compared with those who had always been in the nonmanual class.

DISCUSSION

Women who had experienced socioeconomic adversity throughout their lives were less likely than women who had not to eat healthily and were more likely to have smoked regularly at some point in their lives, to currently smoke, and to be inactive. Our data suggest that both childhood and adult SES affect fruit and vegetable consumption in old age, with roughly an equal strength of association. However, it was primarily adult SES that influenced whether these women were more likely to eat meat or fish. Adult SES appeared to determine quantity and duration of smoking through the age of smoking cessation. Although exercise behaviors in old age were influenced by childhood SES, the effect of adult SES was greater. Upward social mobility and downward social mobility were, respectively, beneficial and detrimental with respect to health behaviors.

Diet

Elderly people come from a generation in which childhood diets were generally healthier in terms of lower saturated fat and calorie content than today. Indeed, it has been reported that older people continue to have better diets; however, deprivation, which may particularly affect elderly people, may partially counteract this trend.16 Other studies have shown that elderly people often have a poor diet that is low in energy and in the amounts of vitamins and minerals consumed. 17-19

Given that few people do so, it is not surprising that small numbers of women in this cohort reported eating the recommended 5 portions of fruits and vegetables per day. Our findings with respect to deprivation are consistent with those of other research.20 People in lower SES groups are more likely to live in areas lacking access to high-quality produce, especially if they do not have access to a car or suffer from poor personal mobility.21Access to social support (e.g., "meals on wheels" programs) may provide a partial solution for the most vulnerable groups.

Childhood SES indicator variables were associated with diet quality in adulthood; however, after adjustment for adult SES, the associations were weaker than the associations of adult SES variables adjusted for childhood SES. This finding suggests that some of the childhood risk factors assessed were mediated through adult SES. The direct effect of childhood SES on diet in old age may result from the tastes developed and the cooking skills and practices learned in childhood.22 This might explain some of the observed effect of childhood SES on adult CHD risk.

Childhood diet may also have a more direct effect on adult health, in that growth affects later disease risk. Leg length (indicative of prepubertal nutritional status) is positively associated with a reduced risk of cardiovascular disease in later life.23,24 Similarly, reduced energy intake in childhood is associated with reduced adult cancer risk.25,26 This direct effect on health may have unforeseen consequences for the increasing number of overweight children in our population.

Smoking

Women who experienced lifelong deprivation were more likely to have smoked in the past and to currently smoke. Among those who had quit, more deprived women generally had smoked for longer periods. These findings are consistent with those of other research on SES and smoking habits.27-29

Our analysis of individual SES indicators suggested that, consistent with other research, adult SES had a strong influence on smoking habits.30 If women perceive more immediate threats to their health (e.g., occupational hazards, street crime), they may downplay the health dangers of smoking and place less priority on stopping smoking.31,32 There is little evidence in our data that childhood SES was related to smoking. This may reflect the attitude toward smoking in the 1940s and 1950s; when these women were young, smoking was more acceptable.

Physical Activity

Most participants were inactive, which is a concern given the strong evidence linking activity with healthy survival in old age.33,34 However, the percentage of women who did exercise for more than 3 hours per week (28%) was higher than the percentages reported among other UK cohorts (e.g., 13% in the English Longitudinal Study of Ageing35). According to the Allied Dunbar National Fitness Survey, conducted in1990, 40%of women aged 65 to 74 years (comparable to the age range in our cohort) reported no physical activity in the previous 4 weeks, and the average was less than 3.5 hours during a 4-week period.36 However, that survey showed that neither education nor social class had an effect on women's exercise behavior.

Our data show that both adult SES and childhood SES were associated with exercise patterns. Few studies have examined the relationship between SES and physical activity across the life span.37 Participation in sports in adolescence is reportedly a predictor of adult physical activity,38 and teenagers in low-SES groups have been shown to be less likely to participate than teenagers in high-SES groups. Other studies have revealed little association between childhood SES and adult physical activity.39 In a separate study involving the present cohort,40 we also found an independent effect of area-level deprivation over and above individual SES, and this is a further and important dimension for consideration in developing health and social policy.

Adult SES affects exercise behavior both directly, as a result of factors such as financial costs (e.g., gym memberships), and indirectly, given that deprivation is associated with increased disability.41 Women without access to a car reported less physical activity than those who had a vehicle, suggesting that walking does not fully compensate for structured exercise opportunities.

Social Mobility

Our data suggest that socially mobile individuals adopt the eating and exercising habits of their new social group. Women whose SES improved over the course of their lives (i.e., women who became more affluent) were more likely than were women whose SES did not improve to eat fruit and vegetables and to exercise; however, they were not as likely to do so as those who had always been members of the nonmanual social class. Conversely, those who moved down the social scale were likely to adopt detrimental health behaviors, but these behaviors were not as harmful, in general, as those engaged in by women who had always been members of the manual social class.

Motivation for some of these behavioral changes may be financial; for example, processed meats are cheaper, and gym admissions and structured exercise programs are expensive. Moreover, many downwardly mobile women married men from poorer backgrounds who then influenced the family's health behaviors.

Strengths and Limitations

Previous studies have highlighted how the use of separate indicators for education, occupation, and family income during childhood adds uniquely to our understanding of how SES is related to behavior.5,6,8 Our work, which extends earlier findings in that we used a much wider range of SES indicators, demonstrates the various ways in which cumulative disadvantage influences health behaviors. Our use of several childhood and adult SES indicators is a strength of this study; it is common practice to use only 1 measure for each, often occupational social class. Adjusting for an individual's socioeconomic position either by conditional logistic regression adjusted for the10-point socioeconomic score, or by using each of the 10 socioeconomic variables as a binary indicator variable did not make any difference to the findings.

We acknowledge that SES measurement error may have influenced the accuracy of our results regarding the independent predictive effects of adulthood and childhood SES on women's health behaviors. However, the modest correlation of 0.33 between SES scores in childhood and adulthood, the differences in the independent predictive power of childhood and adult SES with respect to different health behaviors, and the use of multiple SES indicators at each study time point all suggest that our results are likely to have captured meaningful differences in the predictive power of childhood and adult SES.

Our results were derived from women who were all aged 60 to 79 years at the time of data collection. Without evidence to the contrary, it seems likely that today's children will also retain some of their dietary and exercise habits into adulthood, implying that our findings may have some relevance to the current population of children.

Some women without cardiovascular disease at baseline were excluded from the study because they had missing SES data; however, it is unlikely that exclusion of these women resulted in substantial bias in our analyses, as indicated by the very similar findings obtained in multivariate multiple imputation analyses (data available on request). Dietary data were derived from a simple selfreported food frequency questionnaire that allowed participants to answer questions relatively quickly and easily. The resulting food groups used to indicate a healthful diet were simple but were based on a principalcomponents analysis that produced interpretable groupings. However, our categories did not capture detailed differences in types of food; for example, fish and chicken can include very healthful oily fish and unhealthful fried chicken and fish.

Physical activity as assessed here was not purely a measure of exercise or sporting activities, but rather, included day-to-day activities such as walking and gardening, which are recommended as part of adult activity programs. These forms of physical activity were appropriate for women of the age of our study population and captured the activity level currently recommended by the UK government (30 minutes of moderate activity at least 5 days a week).42

Social desirability bias is a potential issue in all observational studies that collect selfreported behavioral data. However, such bias would tend to attenuate any associations and was unlikely to be sufficiently powerful to remove the widely reported differences in cardiovascular disease outcomes either between socioeconomic groups or by selfreported diet, exercise, or smoking behaviors. Whenever possible, we attempted to validate our risk factor data; for example, we found, in a repeated measures analysis of variance, a significant association between quantity of fruits and vegetables consumed and serum vitamin C levels (P<.001), suggesting that reported intake was a valid indicator (data available on request).

Implications

We have demonstrated that childhood SES, independently of adult SES, is associated with aspects of a healthful diet and physical activity. Our results emphasize the importance of establishing good habits during childhood. School meals in England, after strong criticism, 43 are currently being reformed through government programs. These reform efforts may improve the diets of today's generation of children as they mature.

Home economics classes in which children are taught about food preparation and healthful eating may also be helpful. Successfully educating adults to improve their diets will reduce not only their own CHD risk but that of their children. Targeted programs aimed at increasing physical activity in the poorest communities (where activity levels are lowest), through better provision of opportunities for activity in schools, may also help to increase adult activity levels in years to come.

A focus on the individuals who are currently at highest cardiovascular risk is also warranted. Seemingly the most direct way to improve older people's health behaviors would be to tackle their underlying deprivation. According to recent estimates, the minimum income for pensioners in the United Kingdom to maintain a healthful lifestyle is £122.70 ($236.00) per person per week, somewhat more than the minimum pension credit of £109.45 ($210.50) (including additional benefits such as winter fuel).44 One of the consequences of poverty is that dietary decisions are often financial,45 and members of low-SES groups typically choose unhealthful, cheaper foods. A healthful diet for a moderately active couple in which each partner is older than 65 years costs approximately £63.70 ($122.50) per week, yet average spending in the poorest 40% of couples in this age group is just £44.50 ($85.60) per week.44

The small increases in the UK basic pension instituted in the past 2 years have not been adequate to close these gaps. Older widowed women, previously dependent on their partner's income to raise their family, are particularly affected by today's inadequate pension provisions. Additional financial support for our growing elderly population is needed to ensure people's health in old age.

Conclusions

Our findings highlight the adverse effects of socioeconomic inequalities throughout life on behaviors that are known risk factors for cardiovascular disease and other life-threatening conditions. Improving socioeconomic inequalities in health behaviors and, ultimately, in disease outcomes will require development of better interventions, and these interventions will need to be applied across the life course and will need to focus on disadvantaged groups to provide the greatest benefit.

[Reference] » View reference page with links
References
1. Black D, Morris J, Smith C, Townsend P, eds. The Black Report. London, England: Her Majesty's Stationery Office; 1980.
2. Breeze E, Jones DA, Wilkinson P, Latif AM, Bulpitt CJ, Fletcher AE. Association of quality of life in old age in Britain with socioeconomic position: baseline data from a randomised controlled trial. J Epidemiol Community Health. 2004;58:667-673.
3. The National Service Framework for Coronary Heart Disease: Winning the War on Heart Disease. London, England: Dept of Health; 2004.
4. 5 A DAY Made Easy: Just Eat More (Fruit and Vegetables). London, England: Dept of Health; 2004.
5. Galobardes B, Shaw M, Lawlor DA, Lynch JW, Davey Smith G. Indicators of socioeconomic position: part 1. J Epidemiol Community Health. 2006;60:7-12.
6. Galobardes B, Shaw M, Lawlor DA, Lynch JW, Davey Smith G. Indicators of socioeconomic position: part 2. J Epidemiol Community Health. 2006;60:95-101.
7. Davey Smith G, Egger M. Socioeconomic differentials in wealth and health. BMJ. 1993;307:1085-1086.
8. Lawlor DA, Ebrahim S, Davey Smith G. Adverse socioeconomic position across the lifecourse increases coronary heart disease risk cumulatively: findings from the British Women's Heart and Health Study. J Epidemiol Community Health. 2005;59:785-793.
9. Wamala SP, Lynch J, Kaplan GA. Women's exposure to early and later life socioeconomic disadvantage and coronary heart disease risk: the Stockholm Female Coronary Risk Study. Int J Epidemiol. 2001;30:275-284.
10. Naess O, Claussen B, Davey Smith G. Relative impact of childhood and adulthood socioeconomic conditions on cause specific mortality in men. J Epidemiol Community Health. 2004;58:597-598.
11. Lawlor DA, Bedford C, Taylor M, Ebrahim S. Geographical variation in cardiovascular disease, risk factors, and their control in older women: British Women's Heart and Health Study. J Epidemiol Community Health. 2003;57:134-140.
12. Schroeder K, Lawlor DA, Montaner D, Ebrahim S. Self-reported smoking cessation interventions were not associated with quitting in older women. J Clin Epidemiol. 2006;59:622-628.
13. Lawlor DA, Taylor M, Bedford C, Ebrahim S. Is housework good for health? Levels of physical activity and factors associated with activity in elderly women: results from the British Women's Heart and Health Study. J Epidemiol Community Health. 2002;56:473-478.
14. Royston P. Multiple imputation of missing values. Stata J. 2004;4:227-241.
15. Royston P. Multiple imputation of missing values: update. Stata J. 2005;5:188-201.
16. Swan G. Findings from the latest National Diet and Nutrition Survey. Proc Nutr Soc. 2004;63:505-512.
17. Roberts SB, Hajduk CL, Howarth NC, Russell R, McCrory MA. Dietary variety predicts low body mass index and inadequate macronutrient and micronutrient intakes in community-dwelling older adults. J Gerontol A Biol Sci Med Sci. 2005;60:613-621.
18. de Groot CP, van den Broek T, van Staveren W. Energy intake and micronutrient intake in elderly Europeans: seeking the minimum requirement in the SENECA study. Age Ageing. 1999;28:469-474.
19. Bamia C, Orfanos P, Ferrari P, et al. Dietary patterns among older Europeans: the EPIC-Elderly study. Br J Nutr. 2005;94:100-113.
20. Giskes K, Turrell G, van Lenthe FJ, Brug J, Mackenbach JP. A multilevel study of socio-economic inequalities in food choice behaviour and dietary intake among the Dutch population: the GLOBE study. Public Health Nutr. 2006;9:75-83.
21. Zenk SN, Schulz AJ, Israel BA, James SA, Bao S, Wilson ML. Fruit and vegetable access differs by community racial composition and socioeconomic position in Detroit, Michigan. Ethn Dis. 2006;16:275-280.
22. Maynard M, Gunnell D, Ness AR, Abraham L, Bates CJ, Blane D. What influences diet in early old age? Prospective and cross-sectional analyses of the Boyd Orr cohort. Eur J Public Health. 2006;16:316-324.
23. Lawlor DA, Taylor M, Davey Smith G, Gunnell D, Ebrahim S. Associations of components of adult height with coronary heart disease in postmenopausal women: the British Women's Heart and Health Study. Heart. 2004;90:745-749.
24. Gunnell D, Whitley E, Upton MN, McConnachie A, Smith GD, Watt GC. Associations of height, leg length, and lung function with cardiovascular risk factors in the Midspan Family Study. J Epidemiol Community Health. 2003;57:141-146.
25. Frankel S, Gunnell DJ, Peters TJ, Maynard M, Davey Smith G. Childhood energy intake and adult mortality from cancer: the Boyd Orr Cohort Study. BMJ. 1998; 316:499-504.
26. Hursting SD, Perkins SN, Phang JM, Barrett JC. Diet and cancer prevention studies in p53-deficient mice. J Nutr. 2001;131(suppl 11):3092S-3094S.
27. Power C, Graham H, Due P, et al. The contribution of childhood and adult socioeconomic position to adult obesity and smoking behaviour: an international comparison. Int J Epidemiol. 2005;34:335-344.
28. Myint PK, Luben RN, Welch AA, Bingham SA, Wareham NJ, Khaw KT. Effect of age on the relationship of occupational social class with prevalence of modifiable cardiovascular risk factors and cardiovascular diseases: a population-based cross-sectional study from European Prospective Investigation into Cancer-Norfolk (EPICNorfolk). Gerontology. 2006;52:51-58.
29. Laaksonen M, Rahkonen O, Karvonen S, Lahelma E. Socioeconomic status and smoking: analysing inequalities with multiple indicators. Eur J Public Health. 2005;15: 262-269.
30. Blane D, Hart CL, Smith GD, Gillis CR, Hole DJ, Hawthorne VM. Association of cardiovascular disease risk factors with socioeconomic position during childhood and during adulthood. BMJ. 1996;313:1434-1438.
31. Lawlor DA, Frankel S, Shaw M, Ebrahim S, Smith GD. Smoking and ill health: does lay epidemiology explain the failure of smoking cessation programs among deprived populations? Am J Public Health. 2003;93: 266-270.
32. Stewart MJ, Gillis A, Brosky G, et al. Smoking among disadvantaged women: causes and cessation. Can J Nurs Res. 1996;28:41-60.
33. Schnohr P, Lange P, Scharling H, Jensen JS. Longterm physical activity in leisure time and mortality from coronary heart disease, stroke, respiratory diseases, and cancer: the Copenhagen City Heart Study. Eur J Cardiovasc Prev Rehabil. 2006;13:173-179.
34. Manini TM, Everhart JE, Patel KV, et al. Daily activity energy expenditure and mortality among older adults. JAMA. 2006;296:171-179.
35. Marmot M, Banks J, Blundell R, Lessof C, Nazaroo J. Health, Wealth and Lifestyles of the Older Population in England: The 2002 English Longitudinal Study of Ageing. London, England: Institute for Fiscal Studies; 2003.
36. Allied Dunbar National Fitness Survey: A Report on Activity Patterns and Fitness at All Levels. London, England: London Sports Council and Health Education Authority; 1992.
37. Malina RM. Tracking of physical activity and physical fitness across the lifespan. Res Q Exerc Sport. 1996; 67(suppl 3):S48-S57.
38. Tammelin T, Nayha S, Laitinen J, Rintamaki H, Jarvelin MR. Physical activity and social status in adolescence as predictors of physical inactivity in adulthood. Prev Med. 2003;37:375-381.
39. Parsons TJ, Power C, Manor O. Longitudinal physical activity and diet patterns in the 1958 British Birth Cohort. Med Sci Sports Exerc. 2006;38:547-554.
40. Hillsdon M, Lawlor DA, Ebrahim S, Morris JN. Physical activity in older women: associations with area deprivation and with socioeconomic position over the life course: observations in the British Women's Heart and Health Study. J Epidemiol Community Health. 2008;62: 344-350.
41. Ebrahim S, Papacosta O, Wannamethee G, Adamson J. Social inequalities and disability in older men: prospective findings from the British Regional Heart Study. Soc Sci Med. 2004;59:2109-2120.
42. The National Service Framework for Older People. London, England: Dept of Health Publications; 2004.
43. Gould R, Russell J, Barker ME. School lunch menus and 11 to 12 year old children's food choice in three secondary schools in England-are the nutritional standards being met? Appetite. 2006;46:86-92.
44. Morris J, Dangour A, Deeming C, Fletcher A, Wilkinson P. Minimum Income for Healthy Living: Older People. London, England: Age Concern Reports; 2005.
45. Giskes K, Lenthe F, Brug HJ, Mackenbach J. Dietary intakes of adults in the Netherlands by childhood and adulthood socioeconomic position. Eur J Clin Nutr. 2004;58:871-880.

[Author Affiliation]
Hilary C. Watt, MSc, MA, Claire Carson, PhD, Debbie A. Lawlor, PhD, MBChB, MPH, FFPH, Rita Patel, MSc, and Shah Ebrahim, DM, MSc, BMedSci, FRCP, FFPHM

[Author Affiliation]
About the Authors
Hilary C. Watt, Claire Carson, and Shah Ebrahim are with the Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London, England. Debbie A. Lawlor and Rita Patel are with the Department of Social Medicine, University of Bristol, Bristol, England.
Requests for reprints should be sent to Shah Ebrahim, BMedSci, MSc, DM, FRCP, FFPHM, Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, England (e-mail: shah.ebrahim@lshtm.ac.uk).
This article was accepted May 1, 2008.
Contributors
H. C. Watt contributed to developing the study aims and writing the article and undertook and interpreted the statistical analysis. C. Carson contributed to developing the study aims, undertook the literature review, and made major contributions to the drafting of the article. D. A. Lawlor contributed to developing the study aims and design and contributed to the writing of the article. R. Patel contributed to the drafting of the article. S. Ebrahim supervised the writing of the article and contributed to developing the study aims.
Acknowledgments
The British Women's Heart and Health Study was funded by the Department of Health Policy Research Programme and the British Heart Foundation. Debbie A. Lawlor is funded by a UK Department of Health Career Scientist Award.
We thank Carol Bedford, Alison Emerton, Nicola Frecknall, Karen Jones, Mark Taylor, and Katherine Wornell for collecting and entering British Women's Heart and Health Study data.Wethank all of the general practitioners and their staff who supported collection of data for this study and the women who took part in the study.
Note. The views expressed in this article are those of the authors and not necessarily those of any funding agency. No funding agency influenced data analysis or interpretation.
Human Participant Protection
The British Women's Heart and Health Study received local research ethics committee approval from each of the 23 towns in the study and multicenter approval from the London Multi Region Ethics Committee. All participants provided informed consent.

References
Indexing (document details)
Subjects:Cardiovascular%20disease%22)">Cardiovascular disease, Womens health, Older people, Health behavior, Diet
Author(s):Hilary C Watt, Claire Carson, Debbie A Lawlor, Rita Patel, Shah Ebrahim
Author Affiliation:Hilary C. Watt, MSc, MA, Claire Carson, PhD, Debbie A. Lawlor, PhD, MBChB, MPH, FFPH, Rita Patel, MSc, and Shah Ebrahim, DM, MSc, BMedSci, FRCP, FFPHM

About the Authors
Hilary C. Watt, Claire Carson, and Shah Ebrahim are with the Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London, England. Debbie A. Lawlor and Rita Patel are with the Department of Social Medicine, University of Bristol, Bristol, England.
Requests for reprints should be sent to Shah Ebrahim, BMedSci, MSc, DM, FRCP, FFPHM, Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, England (e-mail: shah.ebrahim@lshtm.ac.uk).
This article was accepted May 1, 2008.
Contributors
H. C. Watt contributed to developing the study aims and writing the article and undertook and interpreted the statistical analysis. C. Carson contributed to developing the study aims, undertook the literature review, and made major contributions to the drafting of the article. D. A. Lawlor contributed to developing the study aims and design and contributed to the writing of the article. R. Patel contributed to the drafting of the article. S. Ebrahim supervised the writing of the article and contributed to developing the study aims.
Acknowledgments
The British Women's Heart and Health Study was funded by the Department of Health Policy Research Programme and the British Heart Foundation. Debbie A. Lawlor is funded by a UK Department of Health Career Scientist Award.
We thank Carol Bedford, Alison Emerton, Nicola Frecknall, Karen Jones, Mark Taylor, and Katherine Wornell for collecting and entering British Women's Heart and Health Study data.Wethank all of the general practitioners and their staff who supported collection of data for this study and the women who took part in the study.
Note. The views expressed in this article are those of the authors and not necessarily those of any funding agency. No funding agency influenced data analysis or interpretation.
Human Participant Protection
The British Women's Heart and Health Study received local research ethics committee approval from each of the 23 towns in the study and multicenter approval from the London Multi Region Ethics Committee. All participants provided informed consent.
Document types:Feature
Document features:Tables, References
Section:RESEARCH AND PRACTICE
Publication title:American Journal of Public Health. Washington: Feb 2009. Vol. 99, Iss. 2; pg. 320, 8 pgs
Source type:Periodical
ISSN:00900036
ProQuest document ID:1630875651
Text Word Count6054
Document URL:http://proquest.umi.com/pqdweb?did=1630875651&sid=1&Fmt=3&clientId=45625&RQT=309&VName=PQD