Empirical Studies About Attendance at Religious Services and Health.

Abstract

In this issue of JAMA Internal Medicine, Li et al1 report a clear and moderately strong association between attendance at religious services and decreased mortality during a 16-year follow-up of a subgroup from the Nurses’ Health Study. The study by Li et al1 includes baseline and follow-up data from 74 534 participants and documented 13 537 deaths. The inverse association of attendance at religious services and adverse health outcomes has been studied extensively, with most results in concert with the study by Li et al, so the results are primarily confirmatory. However, this study is a major contribution to the literature.2 A particular strength of the study, in addition to the large sample size and excellent participation over time by the enrolled women, is the ability to test the temporal association between the independent variable and the outcome variable at multiple time points as well as the use of timevarying covariates to control for confounding, especially by social support and functional status. A favorable distribution of participants across the different categories of attendance at religious services, from almost never to more than once per week, provides ample numbers for comparison. Even so, the study exhibits limitations in terms of generalizability, many of which are acknowledged by the authors yet should be highlighted. Before addressing these limitations, however, we may ask, “What is the rationale for publishing studies about religion in a medical journal focused on documenting empirical evidence related to health and health care?” First, readers and investigators must, as do these authors, focus on the data, no more and no less, and not attempt to generalize beyond the evidence. The study by Li et aldoesnotaddressphilosophicalor theologicalquestions such as, “DoesGod (oranyhigherbeing)exist?”Thedatadonotvalidate claims made about some of the positive benefits of specific religious experiences, claims made even by medical professionals.3 Nor do the data address a biological mechanism by which the religion or spirituality variable enhances health, as do Miller et al4 in their study of the importance of spiritual experience in protecting against the onset of depression, with cortical thickening being associatedwith spirituality in certain regionsof thebrain, suggestingapossiblemechanism.Finally, thedatadonotsuggest thatmedicalprofessionals should recommend attendance at religious services. In other words, the data cannot be taken even as proof of concept for intervention. For such an intervention to be validated, a randomized clinical trial would be required, which is almost certainly unethical, as emphasized by the authors.1 Second, readersmust recognize that studiesof religionand spiritualityhaveproliferateddramatically for thepast 20 to30 years.2 Investigators have answered this question positively, given the significant increase inpublications exploring the association of religion or spirituality and health that have entered the mainstream of scientific reports. Therefore, such studies shouldbeevaluatedusing the samecriteriawithwhich any published empirical study are evaluated. Despite the obvious strengths of the study by Li et al, there are clear limitations,whicharebasicallyembedded in thenatureof thesample itself andfacedbyall investigatorswhoperformsecondarydata analysis. These limitations, inmyview,primarily constrict our ability to generalize from the data presented to the population in general. What are these limitations? The study addresses only one aspect of religion and spirituality, namely, attendance at religious services. Reasons for attendance at religious services may vary appreciably across individuals, such as religious devotion, lifelong habits, social pressures, and perhaps simple loneliness causing individuals to search for a support group with which to connect. One of the strengths of the study is that the investigators explore extensively the role of social supports as a confounding variable in their longitudinal analyses and the explanatory power of attendanceat religious services remains robust in thesecontrolled analyses. However, we have no assurance that attendance at religious services is amarker of the strength of one’s religion or spiritualty and no description of the extent of private practices of spirituality, such as prayer, or perceptions of spiritual well-being among the participants. In addition, the sample is derived from female nurseswho volunteeredtoparticipate in thestudy.Thesewomenare thereforebetter educated than thegeneral population,morewilling to participate in activities that are of value to the larger communitygiventheirvolunteer status, and informedabouthealth and health care in general. The mean baseline age of the participants is 60years or older and therefore the study cannot be Related article page 777 Religious Service Attendance andMortality AmongWomen Original Investigation Research