This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation

Limiting Comorbid Conditions and Breast Cancer Stage at Diagnosis

^a Division of Public Health Biology and Epidemiology, School of Public Health, University of California at Berkeley

Patrice A.C. Vaeth, Dallas Heart Disease Prevention Project, H8.116, University of Texas, Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd., Dallas, TX 75390-9034 E-mail: patrice.vaeth{at}utsouthwestern.edu.

William B. Ershler, MD

	Abstract

Top Abstract Methods Results Discussion References

 
Background. Contrasting hypotheses exist regarding the relationship between comorbidity and breast cancer stage at diagnosis. One suggestion is that disabling comorbid conditions would result in a later stage diagnosis of breast cancer because such conditions would limit mobility and thus access to medical care. This article examines this hypothesis by building a comorbidity summary measure of functionally limiting comorbid conditions and by testing the effectiveness of this measure in predicting the stage at which breast cancer is diagnosed.

Methods. Cases with newly diagnosed breast cancer were identified through the population-based Metropolitan Detroit Cancer Surveillance System, a participant of the Surveillance, Epidemiology, and End Results Program of the National Cancer Institute. Of 1191 eligible cases, 1011 (85%) were interviewed 2–4 months following diagnosis. The analyses for this study were limited to 731 cases for which there were complete data on all variables. Five individual comorbid conditions that predicted functional limitation were combined into a comorbidity summary measure: arthritis, eye conditions, gastrointestinal conditions, kidney conditions, and respiratory conditions. Breast cancer stage was categorized in relation to whether women had local or advanced (regional or remote) disease.

Results. Women with two or more of these five functionally limiting conditions were about half as likely as those with none of these conditions to receive an advanced stage diagnosis of breast cancer (odds ratio [OR] = 0.49, 95% confidence interval [CI] 0.28–0.86, p = .01).

Conclusions. These findings do not support the suggestion that the presence of disabling comorbid conditions results in later stage breast cancer. The five conditions summarized by this measure, although functionally limiting, may also require greater medical monitoring due to associated symptoms and/or treatment requirements and thus lead to increased opportunities for cancer screening.

ONE of the most important determinants of survival from breast cancer is the stage at which it is diagnosed ⁽¹⁾. The presence of concurrent health conditions, or comorbidity, has important implications for breast cancer screening, especially for older women who often have multiple health problems ⁽²⁾. There are several hypotheses regarding how comorbidity affects cancer stage at diagnosis. For instance, comorbidity would be associated with diagnosis of cancer at later stages if existing health conditions caused early cancer symptoms to be overlooked by physicians ⁽³⁾⁽⁴⁾. In addition, disabling conditions, such as stroke or Parkinson's disease, may lead to a later stage breast cancer diagnosis because the mobility limiting nature of such conditions may act as a barrier to accessing medical care ⁽²⁾. A contrasting hypothesis is that comorbidity may result in increased frequency of medical visits and therefore lead to a greater opportunity for early cancer diagnosis ⁽⁵⁾⁽⁶⁾.

There is some evidence to support the latter hypothesis. For example, West and colleagues ⁽⁷⁾ found that in the San Francisco Bay Area women with higher levels of comorbidity were more likely to have been diagnosed with localized breast cancer. Moritz and Satariano ⁽⁸⁾ and Satariano and Ragland ⁽⁹⁾ found a similar marginally significant relationship among women in the Detroit area. However, Havlik and coworkers ⁽¹⁰⁾ and Samet and colleagues ⁽¹¹⁾ found no relationship between cancer stage and comorbidity. The inconsistent findings between comorbidity and disease stage may be due to the use of summary measures of comorbidity that combine several disease types. Summary measures that simply total several diseases may not effectively assess disease stage because different health conditions may affect the timing of breast cancer diagnosis in different ways ⁽²⁾.

Using functional status as an indicator of disability, this article explores the hypothesis that the presence of disabling comorbid conditions leads to later stage breast cancer diagnosis ⁽²⁾. The objectives of this research were to combine those comorbid conditions that best predicted functional limitation into a comorbidity summary measure and to determine the effectiveness of this measure in predicting the stage at which breast cancer is diagnosed.

	Methods

Top Abstract Methods Results Discussion References

 
This article uses information from the Health and Functioning in Women with Breast Cancer study. Begun in 1984, this case-control study examined the functional and psychosocial status of women following the diagnosis and treatment of breast cancer ⁽⁹⁾.

The cases in this study were selected from among all 40–84-year-old women with newly diagnosed breast cancer who resided in the Detroit metropolitan area. All cases were identified through the population-based Metropolitan Detroit Cancer Surveillance System (MDCSS), a participant of the Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute (NCI). Using a rapid reporting system, cases were identified within 4 weeks of diagnosis, which allowed cases to be interviewed from 2–4 months following diagnosis. Women living in long-term care were excluded from the study, as were women with a previous history of breast cancer; however, those with a previous history of other types of cancer were included.

The sample consists of two cohorts identified during two 7-month periods: from 1984 to 1985 and from 1987 to 1988. In the first case series, 571 women, aged 55–84 years, were identified as eligible for study inclusion. Of these, 463 (81.1%) were interviewed. In the second case series, 548 of 620 eligible women (88.4%), aged 40–54 and 75–84 years, were interviewed. The second case series augmented the first by including younger women and by increasing the representation of older women within the sample. In total, 1011 cases were interviewed.

The majority of interviews were conducted in respondents' homes by trained interviewers (5% were performed by phone). The interviews required 45–60 minutes to complete and included questions on sociodemographics, health status, health behavior, functional status, social networks, and social support.

Assessment of Stage of Disease
Information on stage of disease was obtained from the MDCSS and was defined in relation to the criteria established by the NCI's SEER program ⁽¹²⁾. Local disease was diagnosed among cases in which the primary tumor was limited to the breast, nipple, or areola. Cases with regional disease had tumors that had invaded the subcutaneous tissue or skin of the breast, including invasion of, or fixation to, the chest wall, ribs, or chest muscles. Cases with tumors of any size that involved edema or inflammation of the skin or with tumors that had invaded the low, middle, or high axillary lymph nodes were also diagnosed with regional disease. Remote disease is described as metastasized disease, which may involve spread to the skin over the sternum, upper abdomen, axilla, or opposite breast, adrenal gland, or ovaries. Remote disease may also involve invasion of distant lymph nodes, including infraclavicular, supraclavicular, cervical not otherwise specified contralateral axillary, and/or internal mammary nodes, as well as brain, bone, liver, or lung.

Because of the small number of remote cases (n = 36), they were combined with regional stage cases to create an "advanced" stage category. Stage was therefore treated as a dichotomy, with local stage coded as 0 and advanced stage as 1.

According to NCI's SEER program, the 5-year survival rates for local, regional, and remote stage disease are 93%, 72%, and 18% respectively ⁽¹⁾. In 1983–1987, 53% of all breast cancer cases were diagnosed with localized disease, 36% were diagnosed with regional stage disease, and 7% were diagnosed with remote stage disease ⁽¹⁾.

Assessment of Comorbidity
Medical abstractors from the MDCSS reviewed the records of all cases for the presence of 18 concurrent health conditions (hypertension, myocardial infarction, other heart disease, diabetes, arthritis, respiratory disease, stroke, other cancers, broken bones/skeletal disorders, and conditions of the circulatory system, ear, eye, gallbladder, gastrointestinal system, kidney, liver, thyroid, and urinary tract). Abstractors recorded whether each condition was diagnosed prior to, concurrently with, or subsequent to the breast cancer diagnosis. The majority of conditions (93%) were diagnosed prior to the breast cancer diagnosis, and it is these on which the analyses are based. To maintain reliability, the medical records were reabstracted by supervisory staff members and compared with the originals. Inconsistencies, which were subsequently resolved, were found in less than 5% of the cases ⁽⁹⁾. Each condition was coded as 0 for absent and 1 for present.

Assessment of Functional Status
A scale composed of items from questionnaires used in the Massachusetts Health Care Panel Study ⁽¹³⁾, the Framingham Disability Study ⁽¹⁴⁾, and the National Institute on Aging's Established Populations for the Epidemiological Study of the Elderly ⁽¹⁵⁾ was used to assess physical functioning. The scale measured the degree of difficulty respondents reported in carrying out ten basic activities 1 year prior to the interview (i.e., pushing or pulling large objects; stooping, crouching, or kneeling; lifting or carrying items under 10 lbs; lifting or carrying items over 10 lbs; reaching or extending arms above or below shoulder level; writing or handling small objects; standing in place for 15 minutes or longer; sitting for long periods; walking alone up and down stairs; and walking half a mile without help). Respondents indicated their level of difficulty as follows: none, a little, some, a lot, or that they did not perform the activity because of doctor's orders. Each item was coded dichotomously as 0 (no, a little, or some difficulty) and 1 (a lot of difficulty, or didn't perform because of a doctor's orders). Those with a lot of difficulty or who did not perform the activity because of a doctor's orders were considered to have functional limitations, whereas those with no, a little, or some difficulty were considered to have no or little limitation ⁽¹⁶⁾. Responses for the ten items were then added to determine the number of areas in which respondents were functionally limited, and finally, categorized for analyses in relation to whether cases experienced no or little limitation versus limitation in one or more areas.

Assessment of Sociodemographic, Behavioral, and Social Factors
Data on potential confounding variables were obtained from the interviews. Age at the time of the interview was categorized by age group: 40–54, 55–64, 65–74, and 75–84 years. Race was either white or black. Education was coded as a categorical variable: 0–8, 9–11, 12, and 13 or more years of education. A variable to assess financial adequacy 1 year prior to the interview was created from two questions regarding the adequacy of financial resources in meeting daily needs in the past month (adequate, not adequate) and how this compared with 1 year earlier (better, the same, or worse this year). The financial adequacy variable therefore represented the 731 cases whose financial status at the time of the interview reflected that of 1 year earlier. The 230 cases whose prior financial status could not be determined were excluded from the analyses. These women were more likely to be younger, smokers, separated or divorced, and to have no comorbid conditions in comparison with the 731 women for which there was financial adequacy information. This variable was coded in relation to whether resources were adequate or not adequate.

Smoking status 1 year prior to the interview was categorized as follows: nonsmokers who never smoked, former smokers who quit more than 1 year earlier, and current smokers 1 year earlier. Drinking patterns 1 year prior to the interview were assessed from questions regarding the typical frequency and quantity of wine, beer, and other alcohol consumption 1 year earlier. Categories of consumption included: abstainers (did not drink 1 year prior to the interview); infrequent drinkers (drank <1 time/week and consumed at least 1 or 2, but as many as 5 or more drinks/occasion); and frequent drinkers (drank 1 to 2 or more times/week and consumed at least 1 or 2, but as many as 5 or more drinks/occasion) ⁽¹⁷⁾. Body weight was assessed through body mass index (BMI) (i.e., self-reported weight in kg/self-reported height in m²). BMI was categorized in relation to whether cases were not overweight (BMI <25.0), overweight (BMI 25.0–30.0), and obese (BMI >30.0) ⁽¹⁸⁾.

The living arrangement of cases was assessed from questions on marital status, the health of one's husband, and household size. Cases were categorized as living with a healthy spouse, living with an ill spouse and perhaps others, living alone, and living with others ⁽¹⁹⁾. Network size was determined by the number of relatives and friends to whom respondents felt close: 0–2 versus 3–13 friends and relatives. Group membership was indicated by cases' participation in zero versus one or more community, social or recreational, church, self-improvement, or other types of groups.

Comorbidity Summary Measure
The comorbidity summary measure was developed by building a logistic regression model that included all individual comorbid conditions and control variables previously found in chi-square analysis to predict functional limitation with p values of .25 (Table 1 ) ⁽²⁰⁾. The comorbidity summary measure included only those comorbid conditions that were found in the logistic regression analysis to independently predict functional limitation ( p < .10). This measure therefore included the following five conditions: arthritis, gastrointestinal conditions, eye conditions, respiratory conditions, and kidney conditions. Following the strategy used in a measure generated in a previous study of comorbidity and breast cancer survival ⁽⁹⁾, each condition was given a weight of 1. The comorbidity measure therefore consisted of the summed scores of the five conditions and was ultimately coded in relation to whether respondents had zero, one, or two or more of these functionally limiting conditions.

The multivariate analysis was conducted in two stages. First, an initial logistic regression model was built by including the comorbidity summary measure, as well as all other variables previously found in chi-square analysis to have levels of significance of .25 ⁽²⁰⁾. Second, a final model was built that included only variables found in the first model to predict disease stage with a level of significance of .10. A number of possible interactions were also examined (i.e., between the comorbidity measure and age, race, and financial adequacy), but none were found to significantly improve the fit of the model.

	Results

Top Abstract Methods Results Discussion References

 
Sample Characteristics
More than half of the women in the sample were diagnosed with localized breast cancer (Table 2 ), which is similar to the proportion of localized cases reported by the national SEER program ⁽¹⁾. One third of women in the sample reported functional limitations in one or more areas 1 year prior to the interview. The most common area in which cases reported limitation was in stooping, crouching, or kneeling, with 104 women (14.2%) reporting a lot of difficulty, or that they did not perform the task because of a doctor's orders. Ranking second in prevalence were limitations in the following two areas: walking 1/2 mile without help and pushing or pulling large objects, with 101 women (13.8%) reporting limitations in each area. The area in which cases reported the least limitations was in writing or handling small objects, with 22 women (3.0%) reporting limitations (data not shown).

Prevalence of Comorbidity
Review of the cases' medical records for the 18 comorbid conditions showed that the most prevalent condition was high blood pressure, followed by heart disease and arthritis. The least prevalent health problem was liver condition (Table 3 ).

Also at decreased risk of presenting with advanced disease were women who participated in outside groups (OR = 0.70, 95% CI 0.51–0.96, p = .03) and women who lived alone, in contrast with women who lived with a healthy spouse (OR = 0.66, 95% CI 0.43–1.01, p = .06). An increased risk of presenting with advanced stage breast cancer was found among black women, in comparison with white women (OR = 1.55, 95% CI 1.01–2.39, p = .04). When compared with infrequent drinkers, women who drank frequently were also at increased risk of being diagnosed with advanced disease (OR = 1.61, 95% CI 1.06–2.46, p = .03).

Final model: The final logistic regression model included the comorbidity summary measure, as well as age, race, drinking patterns, living arrangement, and group membership. In the final model, the comorbidity summary measure continued to be associated with disease stage at diagnosis (Table 6 ). Those cases with two or more of the five functionally limiting conditions were about half as likely as those with none of these conditions to receive a diagnosis of advanced stage breast cancer.

	Discussion

Top Abstract Methods Results Discussion References

 
The objectives of this analysis were to build a summary measure of functionally limiting comorbidities and to test the effectiveness of this measure in predicting breast cancer stage at diagnosis. Five individual comorbid conditions were found to predict self-reported functional limitation: arthritis, eye conditions, gastrointestinal conditions, kidney conditions, and respiratory conditions. When combined into a comorbidity summary measure, these analyses showed that women with functionally limiting comorbid conditions were at decreased risk of presenting with later stage breast cancer. Women with two or more of these five conditions were, in fact, about half as likely as those with none of these conditions to receive an advanced stage diagnosis.

These findings are in contrast to the suggestion that more disabling conditions may be associated with diagnosis of breast cancer at later stages. It is possible, however, that different comorbidities may affect stage at diagnosis in different ways, with conditions that require medical monitoring leading to early stage diagnosis ⁽²⁾. The five conditions combined in this measure, although functionally limiting, may also be associated with increased medical monitoring due to patients' symptoms and/or treatment requirements. In other words, increased monitoring may occur despite limitations in mobility.

The assessment of functional status was based on the self-reported ability to perform ten tasks. Yet these tasks may not fully reflect one's level of disability. Comorbidities that affect the ability to perform the activities of daily living or instrumental activities of daily living may place more limitations on mobility or in the ability to access medical care, and therefore, may be associated with a later stage diagnosis of breast cancer.

An important strength of this study is its use of population-based cases obtained from the MDCSS, part of the national SEER program. A possible limitation, however, is that these data were collected between 1984 and 1988. Because these data are now more than 10 years old, they may not represent women diagnosed with breast cancer today. It is recommended, therefore, that the association between functionally limiting comorbidities and stage of disease be examined among new cohorts of women in different geographic areas with newly diagnosed breast cancer.

Like other cross-sectional studies, this study is also limited by problems with recall bias and temporal ordering. Recall bias is a concern because information on functional status was obtained at the time of the interview, approximately 3 months after cases had received their diagnosis of breast cancer. Cases' recollection of their functional status 1 year previously may have been influenced by the diagnosis of and treatment for breast cancer. For example, women treated for breast cancer have been found to report greater functional problems than healthy controls ⁽¹⁶⁾ and, as a result, they may believe that their functioning 1 year earlier was better than it actually was.

An additional concern is with temporal ordering. Although analyses were limited to those comorbid conditions that were diagnosed prior to the breast cancer diagnosis, the exact timing of the diagnoses of these comorbidities is unknown. In the analysis, comorbidities were examined as predictors of functional limitations. It is possible, however, that the functional decline preceded the diagnosis of the comorbidities.

Finally, this study did not include information on breast cancer screening behavior. Screening, particularly mammography in combination with clinical breast examination, has been shown to effectively detect breast cancer at earlier stages ⁽²²⁾. Although there are research findings that indicate that comorbidity and older patient age have a negative impact on physicians' attitudes and practices regarding clinical breast examination and mammography referral ^(23)(24)(25), the present study showed that the presence of functionally limiting comorbidity actually increased the likelihood of breast cancer being detected at an earlier stage.

These findings confirm those of others that showed that women with increased comorbidity were more likely to be diagnosed with localized breast cancer ^(7)(8)(9). The findings suggest that increased medical monitoring, regardless of its purpose, is associated with an increased likelihood of early stage breast cancer.

Future research should continue to focus on the relationship between comorbidity and stage of disease at diagnosis. It may be necessary to examine this association in populations that are sufficiently large to ensure that there are adequate numbers of cases in each of the three disease stages: local, regional, and remote. As noted previously, the results from the Detroit study indicated that there was a modest association between level of comorbidity and the likelihood of being diagnosed at a local stage ⁽⁹⁾. Interestingly, there was also evidence of a slight increase in the percentage of women with remote disease with increasing levels of comorbidity; but, because women with remote disease account for a small percentage of women with breast cancer, this pattern did not have a significant effect on the overall relationship. It is important to note, however, that the same general results were found in a study of breast cancer among female members of a health maintenance organization in the San Francisco Bay Area ⁽⁷⁾. With increasing levels of comorbidity, the more likely women were to be diagnosed with either localized or remote disease. There are two possible explanations. First, the association may depend on the relative access of the person to health screening. Those with comorbid conditions who have regular access to health care may be more likely to be diagnosed with localized disease. On the other hand, those with comorbid conditions who do not have such access may be more likely to be diagnosed with remote disease. Second, the association may depend more on the specific combinations of comorbid conditions than on the absolute number.

In the present study, it was not possible to examine which of the individual combinations of comorbid conditions were associated with stage of disease. The frequencies of cases with various combinations of conditions were quite small. Future research needs to be conducted with sample sizes large enough to ensure that there are enough cases with various combinations of conditions. Finally, future research should examine in more detail the overall relationship between comorbidity, stage of disease, and subsequent survival.

	Acknowledgments

 
This research was supported by the National Institute on Aging (Grant RO1-AG04969), a contract from the National Cancer Institute (N01 CN-55423), and the American Cancer Society (Grant PBR-67).

Received May 14, 1999

Accepted December 27, 1999

	References

Top Abstract Methods Results Discussion References

 

1. Miller BA, Ries LAG, Hankey BF, et al. SEER Cancer: Statistics Review, 1973–1990. Bethesda, MD: National Cancer Institute; 1993. National Institutes of Health publication no. 93-2789.
2. Satariano WA, 1992. Comorbidity and functional status in older women with breast cancer: implications for screening, treatment, and prognosis. J Gerontol Med Sci. 47:M24-M31.
3. Steel K, 1983. Caring for the elderly person: focus on cancer. Yancik R, Carbone PP, Patterson WB, et al. , ed.Perspectives on Prevention and Treatment of Cancer in the Elderly 51-61. Raven Press, New York.
4. Crawford J, Cohen HJ, 1984. Aging and neoplasia. Ann Rev Gerontol. 3:3-29.
5. Mor V. Malignant disease and the elderly. In: Evered D, Whelan J, eds., Research and the Ageing Population. Symposium on research and the ageing population, Bethesda, MD, April 28–30, 1987. Chichester, UK: John Wiley and Sons; 1988:160–176.
6. Feinstein AR, 1970. The pre-therapeutic classification of co-morbidity in chronic disease. J Chron Dis. 23:455-469.
7. West DW, Satariano WA, Ragland DR, Hiatt RA, 1996. Comorbidity and breast cancer survival: a comparison between black and white women. Ann Epidemiol. 6:413-419. [Medline]
8. Moritz DJ, Satariano WA, 1993. Factors predicting stage of breast cancer at diagnosis in middle aged and elderly women: the role of living arrangements. J Clin Epidemiol. 46:443-454. [Medline]
9. Satariano WA, Ragland DR, 1994. The effect of comorbidity on 3-year survival of women with primary breast cancer. Ann Intern Med. 120:104-110. [Abstract/Free Full Text]
10. Havlik RJ, Yancik R, Long S, Ries L, Edwards B, 1994. The National Institute on Aging and the National Cancer Institute SEER collaborative study on comorbidity and early diagnosis of cancer in the elderly. Cancer. 74:2101-2106. [Medline]
11. Samet JM, Hunt WC, Goodwin JS, 1990. Determinants of cancer stage: a population based study of elderly New Mexicans. Cancer. 66:1302-1307. [Medline]
12. Shambaugh EM, Weiss MA, Axtell LM, 1977. Summary Staging Guide for the Cancer Surveillance, Epidemiology, and End Results (SEER) Reporting Program National Institutes of Health;, Bethesda, MD.
13. Branch LG, 1988. Applications of epidemiological research in aging to planning and policy: the Massachusetts Health Care Panel Study. Brody JA, Maddox GL, , ed.Epidemiology and Aging Springer, New York.
14. Jette AM, Branch LG, 1981. Framingham Disability Study: II. Physical disability among the aging. Am J Public Health. 71:1211-1216. [Abstract/Free Full Text]
15. Cornoni-Huntley JC, Foley DJ, Ehite LR, et al. 1985. Epidemiology of disability in the oldest old: methodologic issues and preliminary findings. Milbank Q. 63:350-376.
16. Satariano WA, Ragheb NE, Branch LG, Swanson GM, 1990. Difficulties in physical functioning reported by middle-aged and elderly women with breast cancer: a case-control comparison. J Gerontol Med Sci. 45:M3-M11.
17. Vaeth PAC, Satariano WA, 1998. Alcohol consumption and breast cancer stage at diagnosis. Alcohol Clin Exp Res 4:928-934.
18. Expert Panel on the Identification, Evaluation, and Treatment of Overweight in Adults1998. Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults: executive summary. Am J Clin Nutr 68:899-917. [Medline]
19. Satariano WA, Ragheb NE, 1992. Living arrangements and problems with daily tasks for older women with breast cancer. Schaie KW, Blazer D, House J, , ed.Aging, Health Behaviors and Health Outcomes 101-126. Lawrence Erlbaum Associates, Hillsdale, NJ.
20. Hosmer DW, Lemeshow S, 1989. Applied Logistic Regression John Wiley and Sons;, New York.
21. Statistical Package for the Social Sciences (SPSS). Chicago: SPSS; 1997.
22. Baker LH, 1982. Breast Cancer Detection Demonstration Project: five-year summary report. CA Cancer J Clin 194–225:
23. Radecki SE, Kane RL, Solomon DH, Mendenhall RC, Beck JC, 1988. Are physicians sensitive to the special problems of older patients?. J Am Geriatr Soc. 36:719-725. [Medline]
24. Marwill SL, Freund KM, Barry PP, 1996. Patient factors associated with breast cancer screening among older women. J Am Geriatr Soc 44:1210-1214. [Medline]
25. Grady KE, Lemkau JP, McVay JM, et al. 1996. Clinical decision-making and mammography referral. Prev Med. 24:327-338.

This article has been cited by other articles:

				C. J. Bradley, J. P. Clement, and C. Lin Absence of Cancer Diagnosis and Treatment in Elderly Medicaid-Insured Nursing Home Residents J Natl Cancer Inst, January 2, 2008; 100(1): 21 - 31. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Services Download to citation manager Citing Articles Citing Articles via HighWire PubMed PubMed Citation