HomeLarge Type Edition
HOME ARCHIVE SEARCH TABLE OF CONTENTS
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Services
Right arrow Download to citation manager
PubMed
Right arrow PubMed Citation
The Journals of Gerontology Series A: Biological Sciences and Medical Sciences 61:1319-1323 (2006)
© 2006 The Gerontological Society of America

The Temporal Relationship Between Depression Symptoms and Cognitive Functioning in Older Medical Patients—Prospective or Concurrent?

Ling Han, Jane McCusker, Michal Abrahamowicz, Martin Cole and Radan Capek

Departments of 1 Epidemiology and Biostatistics and 4 Pharmacology and Therapeutics, McGill University, Montreal, Canada.
Departments of 2 Clinical Epidemiology and Community Studies and 3 Psychiatry, St. Mary's Hospital, McGill University, Montreal, Canada.

Address correspondence to Jane McCusker, MD, DrPH, McGill University, St. Mary's Hospital, Department of Clinical Epidemiology and Community Studies, 3830 Lacombe Avenue, Montreal, Québec H3T 1M5. E-mail: jane.mccusker{at}mcgill.ca


   Abstract
Top
 Abstract
Methods
 Results
 Discussion
 References
 
Background. Epidemiological studies remain inconclusive as to whether old age depression is an independent risk factor, a prodrome, or a clinical concomitant of cognitive impairment. The objective of this study, using repeated measures over a 12-month period, was to examine the short-term temporal relationship between depressive symptoms and cognitive impairment.

Methods. Two hundred eighty-one medical inpatients 65 years old or older were followed up with the Hamilton Depression Rating Scale (HDRS) and Mini-Mental State Examination (MMSE) at enrollment and 3, 6, and 12 months later. A repeated-measures mixed linear regression model was used to evaluate the association between HDRS scores and MMSE changes over time and to test competing hypotheses about their temporal sequence.

Results. After adjusting for age, cardiovascular risk, illness severity, baseline physical and cognitive function, and other covariates, a one-point increase in HDRS score (baseline mean ± standard deviation: 14.4 ± 7.4) was associated with a lower MMSE score (–0.03, 95% confidence interval, –0.07 to 0.00) at the same time points, but not with the MMSE at subsequent time points (all p values >.40). There were no statistically significant interactions detected between follow-up time and HDRS scores measured at baseline or during follow-up. These results were confirmed in alternative models using dynamic measures of both HDRS and MMSE changes over each successive follow-up interval.

Conclusions. These findings suggest that the short-term relationship between depression symptoms and cognitive functioning may be concurrent or temporary, rather than prospective or protracted, consistent with the clinical concomitant hypothesis.

ALTHOUGH the coexistence of depression symptoms and cognitive impairment in older persons has long been recognized clinically (1), the temporal relationship between the two conditions was not examined from an epidemiological perspective until relatively recently (2). However, early epidemiological studies (3) were often retrospective or cross-sectional in nature, and hence, were inadequate to address the temporality of the relationship. In the past 10 years, a number of large-scale, community-based cohort studies have been undertaken to address the temporal relationship prospectively (4–14).

At least four hypotheses have been postulated to explain the relationship. First, late-life depression may be an independent risk factor of cognitive decline (12,13), perhaps via the "glucocorticoid cascade" pathway (15), in which the progression of depression pathology may ultimately lead to hippocampus damage and dementia. Second, depression and cognitive decline may result from risk factors common to both disorders, such as vascular diseases (16,17). Third, the relationship may be confounded by short-term situational factors (e.g., acute medical illness or disability)—the clinical concomitant hypothesis (3,4,6,11,16). Fourth, depression may be an early manifestation or prodrome of dementia (7–10,14).

Almost all the prospective studies conducted so far have assessed depression and cognitive outcomes at least 1 year, mostly 2–3 years, apart. In addition, most studies evaluated depression symptoms at baseline only as a constant or enduring predictor of cognitive decline (5,9–13). Studies with measures repeated at relatively frequent intervals (months rather than years) can help to elucidate the relationship by examining whether the association is cross-sectional (consistent with the clinical concomitant hypothesis) or prospective (consistent with the prodrome hypothesis).

The objective of the current study was to examine the potential short-term temporal relationship between depression symptoms and cognitive decline, with a specific focus on depression symptoms as a dynamic, time-varying exposure. We used data from a cohort of older medical inpatients that was assessed for both depression symptoms and cognitive function at 3, 6, and 12 months later, with no or little cognitive impairment at study entry.


   METHODS
Top
 Abstract
 Methods
Results
 Discussion
 References
 
Participants
The participants for this study were selected from the study samples of a randomized controlled trial (RCT) of a geriatric psychiatric care service for major depression and an observational cohort study of 12-month outcomes of depression in older medical inpatients, conducted at two university-affiliated acute care hospitals in Montreal, Canada. The enrollment process of the original study has been described elsewhere (18). In brief, 5283 patients 65 years old or older admitted from the emergency room to the medical services were screened by a research clinician using the Short Portable Mental Status Questionnaire (19). Of these patients, 3597 were excluded due to severe cognitive impairment (n = 612, 11.6%) or other reasons (too sick, entered intensive or long-term care, language barriers, or residing outside of Montreal island). The remaining 1686 patients who scored four or less (indicative of no or mild cognitive impairment) were then screened for depression using the depressive disorders section of the Diagnostic Interview Schedule (DIS) (20). Among them, 530 (31.4%) consented to participate in the study. The study protocol was approved by the research ethics committees of both hospitals. Of the 530 enrollees, 22 died and 94 withdrew before the baseline interview, leaving 414 (78.1%) for baseline and follow-up interviews. For this longitudinal analysis, we selected 281 participants with at least two Mini-Mental State Examination (MMSE) scores, representing 67.9% of the baseline cohort of 414. There were no statistically significant differences (all p >.07) between those included (n = 281) and excluded (n = 133) with respect to age, sex, living condition, activities of daily living (ADL) scores, study group, hospital sites, diagnosis of depression, and cognitive impairment at screening. However, patients who were excluded were more severely ill (p <.01), had more comorbid conditions (p <.01), higher Hamilton Depression Rating Scale (HDRS) scores (p =.05), and lower MMSE scores (p =.03).

Measurements
Severity of depressive symptoms was measured using the 21-item version of the HDRS (21), the most widely used interviewer-rated scale for monitoring depressive symptoms and signs in intervention studies of depression. Items were rated from 0 to 4, with a higher score indicating more pathology. Cognitive functioning was measured using the MMSE (22) at the same four time points. With a score ranging from 30 (no impairment) to 0 (maximum impairment), the MMSE is the most widely used brief cognitive instrument for screening cognitive impairment or monitoring its progression (23,24). Studies of its psychometric properties show moderate to high levels of short-term test–retest reliability, construct and criterion validity, and adequate responsiveness to cognitive change over time (23,24). The inter-rater reliabilities of the HDRS and MMSE were assessed in a convenience sample of patients at intervals throughout the study period, by using independent simultaneous ratings by two or more raters, including the study psychiatrist (MC). The intra-class correlation coefficients are 1.00 for both the HDRS (n = 28) and MMSE (n = 17).

Because cardiovascular diseases (CVDs) have been associated with both dementia and depression (4,5,25), we defined a binary indicator (high vs low) for CVD risk. Patients were classified as high risk for CVD if they had a diagnosis of stroke, diabetes, or myocardial infarction during the previous 2 years or a measured sitting blood pressure of at least 160/95 mmHg from the hospital chart. Independence in ADL was assessed at baseline by the research assistant with the Older Americans Resources and Services (OARS) ADL scale (26), with a score range from 0 (completely dependent) to 14 (completely independent). History of alcohol abuse was obtained by the research assistant using the four-item informant questionnaire Cut down, Annoyed, Guilty, Eye-opener (CAGE; rated from 0 for no alcohol use to 4 for heavy alcohol use) (27). Age, sex, education, living condition prior to the admission, Charlson Comorbidity Index (CCI), a composite measure by number and severity of comorbid conditions (28), and a nurse-rated illness severity score (from 0 for not ill to 9 for moribound) (29) were obtained either from interview or hospital charts abstraction.

Statistical Analyses
The characteristics of the study population at baseline and the distribution of HDRS and MMSE scores over time were described using means and standard deviations or frequencies and proportions, as appropriate.

We used a mixed effects linear regression model approach to examine the temporal relationship between depression symptoms and cognitive functioning over time, in which the HDRS and MMSE scores were both updated every 3 or 6 months during follow-up. The mixed model allows for both fixed (time-invariant) covariates, the values of which do not change over time (such as sex), and time-dependent covariates, the values of which can be updated during the follow-up (such as HDRS scores) (30). We adopted the Spatial Power covariance structure of errors to account for potential intercorrelations among the repeated measures on the same patient. This structure assumes the correlations between any two measures to decrease as their distance increases, and allows for the unequal follow-up intervals and number of assessments across participants (31).

We tested two sets of operational mixed models under competing hypotheses, termed as "concurrent" and "prospective," respectively, by different representations of HDRS and MMSE scores. In the concurrent model, the HDRS was associated with the MMSE changes at the same follow-up time points, without a clear-cut temporal or causal implication. In the prospective model, the HDRS at each follow-up was used to predict the MMSE changes at the next follow-up after a 3- or 6-month time lag, which allows us to evaluate the depression symptoms as a potential causal risk factor of cognitive decline without temporal ambiguity. In contrast, the last available HDRS score of a patient has to be discarded, which reduces the statistical power to some degree. We also fit a general linear regression model, with baseline HDRS score as a predictor and the difference between baseline and last available MMSE score as an outcome.

Covariates were adjusted in a hierarchical fashion. First, we decided a priori to include age, education, CVD risk, ADL function, hospital sites, study group (RCT-intervention, RCT-control, Non-RCT), and follow-up time in all the models, regardless of their statistical significance, given their established importance in confounding the relationship between depression and cognitive decline in the literature or due to the study design. Second, we tested effects of other covariates, including sex, living arrangement, illness severity, CCI, and CAGE score, individually and simultaneously, but did not retain them in the final models due to lack of statistical significance. In addition, we sequentially adjusted for the baseline MMSE score, attempting to control for potential confounding by unmeasured factors or events that might have operated on the participants' cognition before the start of the follow-up, and the baseline HDRS score, in case the longitudinal effects of depression symptoms may have been predetermined by their initial level. Finally, we tested the interaction between baseline HDRS scores and follow-up time in the final models. If a statistically significant interaction was detected, separate models would be estimated for each follow-up interval. Depression group was excluded from the multivariate regression models because of the substantial conceptual overlap between this variable and HDRS score, and the significant correlation between the two variables (Spearman's {rho} = 0.58, p <.001).

All statistical analyses were conducted using SAS software version 9.1 (31). Goodness of fit of nested models was compared using the Akaike's Information Criterion (AIC) (32). The hypotheses were tested at a two-sided significance level of {alpha} = 0.05.


   RESULTS
Top
 Abstract
 Methods
 Results
Discussion
 References
 
Characteristics of the Study Population
The characteristics of the study population at baseline are presented in Table 1. A total of 61% of the sample was depressed at baseline. The study sample had a mean MMSE score of 25.8 (standard deviation = 3.5) with 26.0% below 24. The mean MMSE scores (25.6 vs 26.3, p =.12) and mean numbers of Short Portable Mental Status Questionnaire errors (1.6 vs 1.6, p =.82) were similar between the depressed and nondepressed patients.


View this table:
[in this window]
[in a new window]

  		Table 1. Characteristics of the Study Population at Baseline.

 
The distribution the HDRS and MMSE scores across time is summarized in Table 2. There was a trend of negative (Pearson product-moment) correlations between the two measures at baseline (r = –0.10, p =.08) and 3 months (r = –0.12, p =.07).


View this table:
[in this window]
[in a new window]

  		Table 2. Distribution of Repeated-Measures HDRS and MMSE Scores During Follow-Up.

 
Table 3 presents the results of a series of mixed regression models. Of the three "concurrent" models, Model 2 has a minimum value of AIC, and thus can be considered as providing best fit to the data. It suggests that for patients with comparable cognitive function at baseline, a 1-point increment in the HDRS score is associated with a decline of –0.03 MMSE point (95% confidence interval [CI]: –0.07 to 0.00, p =.05) when measured at the same follow-up time points, after adjusting for other covariates. This estimate did not change materially when the baseline MMSE was not adjusted for (–0.04, 95% CI: –0.07 to –0.02, p <.01, Model 1) or when an additional covariate, the baseline HDRS, was adjusted (–0.04, 95% CI: –0.08 to –0.01, p =.02, Model 3). On the contrary, none of the three "prospective" models (Models 4–6) yields a statistically significant association between depression symptoms and subsequent cognitive declines (all p values above.40). In either concurrent or prospective models, no statistically significant interaction between depressive symptoms and follow-up time was detected (all p values above.25, data not shown).


View this table:
[in this window]
[in a new window]

  		Table 3. Mixed Linear Regression Models Evaluating the Effects of Depression Symptoms on MMSE Changes* Over Time.

 
In additional mixed-model analyses in which both HDRS and MMSE were represented by their score changes between two adjacent follow-ups, the associations between the two measures were statistically significant only in concurrent models (all p values below.05), not in the prospective models (all p values above.75). Similarly, the multiple linear regression models, in which the time lag between the measures of the depression symptoms and the MMSE changes was extended up to 12 months, failed to detect an independent association (all p values above.35, data not shown). To ensure that our exclusion of the secondary covariates did not introduce bias, we included sex, living arrangement, illness severity, CCI score, and CAGE score altogether in the final models, and found no material changes in the effect estimates for the HDRS scores in both concurrent and prospective models (data not shown).


   DISCUSSION
Top
 Abstract
 Methods
 Results
 Discussion
References
 
In this cohort of 281 older medical inpatients followed up to 12 months, we systematically evaluated two sets of statistical models under alternative hypotheses about the temporal sequence between depression symptoms and cognitive decline. After controlling for the effects of a number of potential confounding factors and initial level of cognitive function, we observed that depression symptoms were independently associated with worse cognitive functioning cross-sectionally, at the same follow-up time points. However, such an association disappeared when the exposure and outcome were separated temporally by a 3- to 6-month time lag, a period corresponding to a typical major depressive episode (33). These results were confirmed in the alternative mixed-effects models using dynamic measures of depression symptoms and cognitive functioning and in the general linear model evaluating the relationship with a maximum follow-up interval of 12 months. Our study suggests that depressive symptoms are a clinical concomitant, rather than a predictor or prodrome of cognitive decline.

Our observation of a concurrent association is consistent with several previous cohort studies in older community-dwelling populations. Dufouil and colleagues (4) followed 1600 elderly persons 65 years old or older in France and found the MMSE scores at 3-year follow-up were only cross-sectionally associated with Center for Epidemiological Studies–Depression Scale (CES-D) scores measured at the same time point, not at baseline. Similarly, Henderson and colleagues (5), Chen and colleagues (8), Cervilla and colleagues (11), and Vinkers and colleagues (14) all failed to find an independent prospective association between depression and subsequent cognitive decline or onset of dementia. However, the follow-up intervals in these studies were much longer than those in this one, ranging from 1 to 12 years.

Several reasons may underlie a cross-sectional or concurrent relationship. First, it may be determined or mediated by a shared short-term risk factor such as acute medical illness or functional disability (3,16,17), a recent stressful life event (e.g., bereavement), or use of antidepressant medications (34). We plan to evaluate these factors in future analyses of this cohort. Alternatively, a cross-sectional relationship may be an artifact, due to poorer performance on cognitive tests like the MMSE among depressed people, especially on the items that demand strong attention, motivation, or psychomotor speed (35).

Our study has several strengths. First, we focused specifically on the short-term effects of depression symptoms as a dynamic, time-varying exposure, using a clinically plausible effect period for depression symptoms (33) and an appropriate longitudinal modeling approach. Second, we carefully selected and rigorously controlled important confounders based on both substantive knowledge and statistical efficiency, including CVD, functional disability, and illness severity. Third, we avoided a "reverse causality" bias by excluding patients with moderate or severe cognitive impairment at study entry and adjusting for baseline cognitive function in the analyses. Finally, we enhanced the study validity by using blind exposure and outcome assessments and by adopting an objective, interviewer-rated (i.e., HDRS) depression scale.

Several study limitations should be noted. First, the MMSE has been criticized for insensitivity to small cognitive changes and ceiling or floor effects (23,24). Second, the HDRS has been criticized for its inclusion of somatic symptoms. These measurement issues may have biased our results towards the null. A third limitation is the relatively high rate of exclusion and cohort attrition, perhaps not surprising in medically ill older people. Given that the excluded patients tended to be more severely ill than those in the study sample, our findings may not be generalizable to the most severely ill older persons or to those outside hospital settings.

Conclusion
We have documented the existence of a concurrent or temporary, rather than a prospective or protracted, association between depression symptoms and cognitive decline in this cohort of older medical patients, independent of other potentially important risk factors. These results do not support the hypothesis that depression symptoms in older people are an independent short-term risk factor for cognitive decline or a prodrome of dementia. Rather, they suggest that the two conditions occur concomitantly. Future studies should account for extraneous factors that may account for this association, such as recent life events and medication use, and improve the measurements for both depression and cognition.


   Acknowledgments
Top
 Abstract
 Methods
 Results
 Discussion
 References
 
This study was funded by Canadian Institutes for Health Research, Grant MOP82494 to Dr. Jane McCusker and Dr. Martin Cole. An abstract was presented at 2006 Congress of Epidemiology, June 21–24, Seattle, Washington, with partial support from the Yale Claude D. Pepper Older Americans Independence Center (NIA grant P30AG021342) for Dr. Ling Han.

We thank Eric Belzile for his assistance in managing and organizing the computerized database, and all the research assistants for recruiting participants and collecting the data.


   Footnotes
Top
 Abstract
 Methods
 Results
 Discussion
 References
 
Decision Editor: Darryl Wieland, PhD, MPH

Received November 12, 2005

Accepted April 28, 2006


   References
Top
 Abstract
 Methods
 Results
 Discussion
 References
 

  1. Meyers BS, Bruce ML. The depression-dementia conundrum. Arch Gen Psychiat. 1998;55:1082-1083.[Free Full Text]
  2. Jorm AF, van Duijn CM, Fratiglioni L, et al. Psychiatric history and related exposures as risk factors for Alzheimer's disease: a collaborative re-analysis of case-control studies. Int J Epidemiol. 1991;20:(suppl 2): s43-s47.
  3. Fuhrer R, Antonucci TC, Gagnon M, Dartigues JF, Barberger-Gateau P, Alpérovitch A. Depressive symptomatology and cognitive functioning: an epidemiological survey in an elderly community sample in France. Psychol Med. 1992;22:159-172.[Medline]
  4. Dufouil C, Fuhrer R, Dartigues JF, Alperovitch A. Longitudinal analysis of the association between depressive symptomatology and cognitive deterioration. Am J Epidemiol. 1996;144:634-641.[Abstract/Free Full Text]
  5. Henderson AS, Korten AE, Jacomb PA, et al. The course of depression in the elderly: a longitudinal community-based study in Australia. Psychol Med. 1997;27:119-129.[Medline]
  6. Devanand DP, Sano M, Tang MX, et al. Depressed mood and the incidence of Alzheimer's disease in the elderly living in the community. Arch Gen Psychiatry. 1996;53:175-182.[Abstract/Free Full Text]
  7. Bassuk SS, Berkman LF, Wypij D. Depressive symptomatology and incident cognitive decline in an elderly community sample. Arch Gen Psychiatry. 1998;55:1073-1081.[Abstract/Free Full Text]
  8. Chen P, Ganguli M, Mulsant BH, DeKosky ST. The temporal relationship between depressive symptoms and dementia: a community-based prospective study. Arch Gen Psychiatry. 1999;56:261-266.[Abstract/Free Full Text]
  9. Yaffe K, Blackwell T, Gore R, Sands L, Reus V, Browner WS. Depressive symptoms and cognitive decline in nondemented elderly women: a prospective study. Arch Gen Psychiatry. 1999;56:425-430.[Abstract/Free Full Text]
  10. Geerlings MI, Schoevers RA, Beekman AT, et al. Depression and risk of cognitive decline and Alzheimer's disease. Results of two prospective community-based studies in The Netherlands. Br J Psychiatry. 2000;176:568-575.[Abstract/Free Full Text]
  11. Cervilla JA, Prince M, Joels S, Mann A. Does depression predict cognitive outcome 9 to 12 years later? Evidence from a prospective study of elderly hypertensives. Psychol Med. 2000;30:1017-1023.[Medline]
  12. Paterniti S, Verdier-Taillefer MH, Dufouil C, Alpérovitch A. Depressive symptoms and cognitive decline in elderly people. Br J Psychiatry. 2002;181:406-410.[Abstract/Free Full Text]
  13. Wilson RS, Barnes LL, de Leon M, et al. Depressive symptoms, cognitive decline, and risk of AD in older persons. Neurology. 2002;59:364-370.[Abstract/Free Full Text]
  14. Vinkers DJ, Gussekloo J, Stek ML, Westendorp RGJ, van der Mast RC. Temporal relation between depression and cognitive impairment in old age: prospective population based study. BMJ. 2004;329:881-884.[Abstract/Free Full Text]
  15. Jorm AF. History of depression as a risk factor for dementia: an updated review. Aust N Z J Psychiatry. 2001;35:776-781.[Medline]
  16. Lavretsky H, Small GW. Mixed cognitive and depressive syndromes. In: Roose SP, Sackeim HA, eds. Late-Life Depression. New York: Oxford University Press; 2004.
  17. Alexopoulos GS, Meyers BS, Young RC, Campbell S, Silbersweig D, Charlson M. "Vascular depression" hypothesis. Arch Gen Psychiatry. 1997;54:915-922.[Abstract/Free Full Text]
  18. McCusker J, Cole M, Dufouil C, et al. The prevalence and correlates of major and minor depression in older medical inpatients. J Am Geriatr Soc. 2005;53:1344-1353.[Medline]
  19. Pfeiffer E. A short portable mental status questionnaire for the assessment of organic brain deficit in elderly patients. J Am Geriatr Soc. 1975;23:433-441.[Medline]
  20. Robins LN, Helzer JE, Croughan J, et al. National Institute of Mental Health Diagnostic Interview Schedule: its history, characteristics, and validity. Arch Gen Psychiatry. 1981;38:381-389.[Abstract/Free Full Text]
  21. Hamilton M. Development of rating scale for primary depressive illness. Br J Soc Clin Psychol. 1967;6:278-296.[Medline]
  22. Folstein MF, Folstein SE, McHugh PR. Mini-Mental State: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12:189-198.[Medline]
  23. Tombaugh TN, McIntyre NJ. The mini-mental state examination: a comprehensive review. J Am Geriatr Soc. 1992;40:922-935.[Medline]
  24. Galasko D, Corey-Bloom J, Thai LJ. Monitoring progression in Alzheimer's disease. J Am Geriatr Soc. 1991;39:932-941.[Medline]
  25. Skoog I. Risk factors for vascular dementia: a review. Dementia. 1994;5:127-144.
  26. Fillenbaum GG. Multidimensional functional assessment of older adults: the Duke Older Americans Resources and Services procedures. Hillsdale, NJ: Lawrence Erlbaum Associates Inc.; 1988:179.
  27. King M. At risk drinking among general practice attenders: validation of the CAGE questionnaire. Psychol Med. 1986;16:213-217.[Medline]
  28. Charlson ME, Pompei P, Ales KL, MacKenzie RC. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373-383.[Medline]
  29. Charlson ME, Sax FL, MacKenzie R, Fields SD, Braham RL, Douglas RG. Assessing illness severity: does clinical judgment work? J Chronic Dis. 1986;39:439-452.[Medline]
  30. Jennrich RI, Schluchter MD. Unbalanced repeated-measures models with structured covariance matrices. Biometrics. 1986;42:805-820.[Medline]
  31. SAS Version 9.1., SAS Institute Inc., Cary, NC; 2003.
  32. Akaike H. A new look at the statistical model identification. IEEE Transaction on Automatic Control. 1974;AC-19:719–723.
  33. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th Ed. Washington, DC: American Psychiatric Association; 1994.
  34. Oxman TE:. Antidepressants and cognitive impairment in the elderly. J Clin Psychiatry. 1996;57:(suppl 5): 38-44.
  35. Boone KB, Lesser LM, Miller B, et al. Cognitive functioning in older depressed outpatients: relationship of presence and severity of depression to neuropsychological test scores. Neuropsychology. 1995;9:390-398.




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Services
Right arrow Download to citation manager
PubMed
Right arrow PubMed Citation

HOME ARCHIVE SEARCH TABLE OF CONTENTS