Elsevier

Social Science & Medicine

Volume 57, Issue 12, December 2003, Pages 2423-2434
Social Science & Medicine

Networked for change? identifying obstetric opinion leaders and assessing their opinions on caesarean delivery

https://doi.org/10.1016/S0277-9536(03)00137-0Get rights and content

Abstract

The objective was to determine whether obstetric opinion leaders can be identified and to characterize them in terms of their demographic and professional characteristics and their attitudes toward caesarean delivery. In late 1998, we surveyed 527 obstetricians, 138 family physicians, and 80 certified nurse midwives (overall response rate, 57.8%) practicing in a stratified random sample of California hospitals with at least 1000 annual deliveries (n=52). Participants reported on demographic and professional characteristics and attitudes towards caesarean delivery; they also checked off those hospital colleagues from whom they had sought or would seek advice on labour and delivery. A composite measure of nomination frequency was used to characterize each respondent's degree of “opinion leadership”. All analyses were corrected for the complex survey design. Using a nomination cutoff of 0.4 (0–1 scale), opinion leaders were identified in 31% of California hospitals; they were identified in 81% of hospitals using a cutoff of 0.2. Compared with their peers in the lowest fifth of the nomination distribution, clinicians in the top fifth were younger and more likely to be male, to speak English as a first language, to practice obstetrics, to have a maternal–foetal medicine subspecialty, and to practice in higher volume hospitals (p<0.05). Regardless of discipline, opinion leaders held attitudes concordant with reducing the caesarean delivery rate more often than non-opinion leaders. However, only 48% of obstetrical opinion leaders would support reducing the caesarean delivery rate to levels targeted by Healthy People 2000. In conclusion, obstetric opinion leaders could be identified in many California hospitals. However, they did not consistently support policies designed to reduce the caesarean delivery rate. The results have implications for the generalizability of opinion leader strategies.

Introduction

Physicians are intimately involved in nearly every aspect of medical decision making. Yet considerable evidence suggests that their decisions are not always evidence based, patient-centred, or cost-effective (Nordin-Johansson & Asplund, 2000; Little et al., 2001; Chapman, Stone, Sandberg, Bell, & Neumann, 2000). Getting physicians to alter their practices could have substantial impact on health care quality and costs. However, the usual approaches to changing physician behaviour have shortcomings. Financial incentives are effective but blunt, and they tend to encourage changes in volume of care but not necessarily appropriateness or quality (Rogers et al., 1990). Traditional forms of continuing medical education have been disappointing (Davis et al., 1999). Practice guidelines are often ignored (Lomas, 1991). Reminders, audit and feedback, and computerized decision aids have proven useful in some settings but not in others (Thomson O’Brien et al., 2000a; Oxman, Thomson, Davis, & Haynes, 1995). The search continues for ways to improve physician practice at a reasonable cost.

One promising approach to changing physician behaviour is the use of local clinical opinion leaders. According to one definition, opinion leaders are health professionals nominated by their colleagues as “educationally influential” (Hiss, MacDonald, & David, 1978). Their utility as change agents is predicted by social influence theory, which posits that clinicians may be “influenced significantly by colleagues’ judgements of the value and significance of [an intervention] and/or by their decisions to use or ignore it” (Mittman, Tonesk, & Jacobson, 1992). In practice, however, use of local opinion leaders has not been uniformly effective: the most comprehensive review to date found that the results were statistically significant and clinically important in only 2 of 8 clinical trials (Thomson O’Brien et al., 2000b).

Locock, Dopson, Chambers, and Gabbay (2001) recently adduced two possible reasons for these mixed results. First, opinion leaders may be difficult to identify, in part because they are not all cut from the same cloth. Some opinion leaders (e.g., “acknowledged experts”) may be more valuable during the introduction of an innovation, whereas others (e.g., “respected peers”) may be more important during the implementation and consolidation phase. Second, opinion leaders may not always support a practice change or innovation and may in fact be hostile to it. Thus, in attempting to discern whether opinion leaders might successfully influence practice in their clinical communities, it is important to know something of their attitudes and opinions.

Obstetric care is a rich area for investigation of the role of opinion leaders. During the late 1980s, a number of professional organizations, consumer groups, health plans, and the federal government launched initiatives designed to combat what was then termed an “epidemic” of caesarean delivery in the United States. Caesarean rates declined in the early 1990s but have again risen steadily since 1997—a trend that is international in scope (Leung, Lam, Thach, Wan, & Ho, 2001; Murray & Serani Pradenas, 1997). Although the clinical issues are controversial, many experts believe that the current US caesarean section rate of 24% could be safely reduced. Decisions about mode of delivery are known to be influenced by social as well as clinical factors (Hurst & Summey, 1984). The organization of obstetric practice in the United States is such that obstetrical clinicians (including obstetricians, family physicians, and nurse midwives) are likely to encounter one another in the hospital, providing ample opportunity for clinical interaction and mutual influence during rounds, conferences, consultations and informal meetings.

Previous studies and case reports have relied primarily on informal methods to identify opinion leaders. For example, in a trial of educational visits to enhance use of systematic reviews in obstetric units, the investigators visited the “lead obstetrician and midwife on the labour ward… because they had usually been nominated to hold these positions by peers as being the most involved in labour ward management, policy making, and training” (Wyatt et al., 1998). In a trial of education and opinion leaders to improve adherence to dementia guidelines, another research group asked neurologists to list up to three local colleagues who were knowledgeable, compassionate, and skilled as teachers (Gifford et al., 1999; Holloway, Gifford, Frankel, & Vickrey, 1999). The neurologists most frequently mentioned in a given region were asked to serve as project opinion leaders.

Sociometry is the use of quantitative approaches to describe relationships within social networks. We employed sociometric techniques to identify local obstetrical opinion leaders in 52 California hospitals. In so doing, we asked three research questions. First, can opinion leaders be identified? If networks of obstetrical providers are so diffuse that clinicians with disproportionate influence cannot be identified in most hospitals, then the generalizability of the opinion-leader approach must be questioned.

Second, what demographic and professional characteristics are associated with high opinion-leader status? We expected that obstetrical opinion leaders would be distinguished by personal and professional characteristics traditionally associated with higher professional status such as (middle) age, (male) gender, and advanced clinical training (i.e., subspecialization). Confirmation of these relationships would support the validity of sociometric methods for identifying opinion leaders. In addition, finding strong associations between clinicians’ outward characteristics and opinion leader status might facilitate efficient searches for opinion leaders in future projects aimed at changing clinician behaviour.

Our third research question was how do the attitudes, beliefs, and opinions of obstetrical opinion leaders differ (if at all) from those of their sociometrically less-esteemed colleagues? Even if obstetrical opinion leaders are identifiable in most hospitals, they will not be useful allies for reducing the current caesarean delivery rate safely and appropriately unless they are positively inclined to do so or can be so persuaded (Coleman et al. (1957), Coleman, Katz, & Menzel (1966); Coleman et al., 1957).

Section snippets

Sampling of hospitals

Using data available from the California Office of Statewide Health Planning and Development (OSHPD), we identified 194 California hospitals where at least 1000 infants were delivered in 1995. The sampling frame was limited to these hospitals because hospitals with very low obstetrical volumes were unlikely to have sufficient obstetrical providers for meaningful network analysis.

The 194 hospitals with at least 1000 deliveries in 1995 accounted for 88.7% of all deliveries in California in that

Respondent characteristics

Seventy-one percent of respondents were obstetrician–gynecologists, 19% were family physicians, and 11% were certified nurse midwives (Table 1). The mean age was 48 years; 35% were female and 29% were non-white. About 19% were not native speakers of English (data not shown in table). A majority (60%) reported working in solo or single-specialty group practice. The average respondent delivered 16 infants per month, of whom 20% were “high risk”. Among obstetricians and family physicians, the

Discussion

Our findings indicate that clinicians nominated by their peers as a source of credible obstetrical advice are identifiable in many hospitals and are distinguishable from their colleagues along plausible demographic and professional dimensions. These obstetric opinion leaders are more favourably inclined towards policies that would reduce the overall c-section rate than are their less influential colleagues. However, they are by no means solidly behind such policies.

Before discussing these

References (43)

  • R.H. Chapman et al.

    A comprehensive league table of cost-utility ratios and a sub-table of “panel-worthy” studies

    Medical Decision Making

    (2000)
  • J. Coleman et al.

    Medical innovationA diffusion study

    (1966)
  • J. Coleman et al.

    The diffusion of an innovation among physicians

    Sociometry

    (1957)
  • D. Davis et al.

    Impact of formal continuing medical educationDo conferences, workshops, rounds, and other traditional continuing education activities change physician behavior or health care outcomes?

    The Journal of the American Medical Association

    (1999)
  • Eltinge, J. L., & Scribney, W. M. (1996). Estimates of linear combinations and hypothesis testing for survey data. In:...
  • M. Fishbein et al.

    Belief, attitude, intention, and behaviorAn introduction to theory and research

    (1975)
  • B.L. Flamm et al.

    Vaginal birth after cesarean deliveryResults of a 5-year multicenter collaborative study

    Obstetrics and Gynecology

    (1990)
  • D.R. Gifford et al.

    Improving adherence to dementia guidelines through education and opinion leaders. A randomized, controlled trial

    Annuals of Internal Medicine

    (1999)
  • R.G. Hiss et al.

    Identification of physician educational influentials in small community hospitals

    Research in Medical Education

    (1978)
  • D. Krackhardt

    The strength of strong tiesThe importance of philos in organizations

  • Krackhardt, D., & Hanson, J. R. (1993). Informal networks: The company behind the chart. Harvard Business Review, July...
  • Cited by (24)

    • Is blood thicker than water? Peer effects in stent utilization among Floridian cardiologists

      2011, Social Science and Medicine
      Citation Excerpt :

      Much research finds the potential influence of other physicians and key opinion leaders remains strong (Bikhchandani, Chandra, Goldman, & Welch, 2002; Berwick, 2003; Borbas, Morris, McLaughlin, Asinger, & Gobel, 2000; Burke, Fournier, & Prasad, 2007; Chandra & Staiger, 2007; Escarce, 1997; Grilli & Lomas, 1994; Greenhalgh, Robert, MacFarlane, Bate, & Kyriakidou, 2004; Meltzer, 2009; Soumerai et al, 1998). Yet the transfer of best practices within healthcare settings has not been simple in the USA and in affluent Western countries more broadly (Doumit, Gattellari, Grimshaw, & O’Brien, 2007; Fattore, Frosini, Salvatore, & Tozzi, 2009; de Jong, Groenewegen, Spreeuwenberg, Schellevis, & Westert, 2010Keating, Ayanian, Cleary, & Marsden, 2007; Kuo, Gifford, & Stein, 1998), and practice patterns appear relatively immutable to the influence of peers (Kravitz et al., 2003; Nicholson & Epstein, 2003; Tucker et al., 2007). Observed variations in the utilization of medical service remain widespread in the USA (Fisher et al., 2003), while unobserved differences in which patients are treated likely mask further variation in community health (Huesch, 2010a).

    • Exploring the use of social network methods in designing healthcare quality improvement teams

      2010, Social Science and Medicine
      Citation Excerpt :

      Building on the analysis of Burt (2005) and other contributors to the management literature on teams (Cross, Ehrlich, Dawson, & Helferich, 2008), this requires choosing individuals based on their connections to persons both within and outside the team. While previous studies in the clinical literature have used social network principles to identify effective single opinion leaders (Kravitz et al., 2003; Soumerai et al., 1998), we are not aware of prior studies that have used SNA to improve the design of quality improvement teams in healthcare, which is our ultimate goal. Our objectives in this paper are twofold.

    View all citing articles on Scopus
    View full text