 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
,
Department of Pediatrics Department of Public Health To the Editor.—
We read with great interest the article by Olds et al1 on home visiting by paraprofessionals and nurses. Many of us across the disciplines are indebted to Dr Olds for the seminal work he and his colleagues have conducted over the last several decades demonstrating the efficacy of his nurse home visitation program in preventing a broad array of detrimental maternal health, life course, and child developmental outcomes. Dr Olds’ work has been especially instrumental in the field of child abuse and neglect prevention, helping to spur the development of hundreds of preventive home visitation programs across the United States.
The findings reported in Pediatrics in September 2002 extend his earlier work by again demonstrating a wide array of positive benefits for mother-child pairs as a result of nurse home visitation, now across 3 different community settings: Elmira, New York, Memphis, Tennessee, and Denver, Colorado. We would like to caution the readers, however, concerning the conclusions they might be tempted to draw from this most recent study, which, on the face of it, appears as if Dr Olds has directly demonstrated the relative efficacy of nurses versus paraprofessionals as home visitors, in a way generalizable to other contexts. Dr Olds and colleagues, themselves, directly caution us, "Because of constraints of sample size and cost, the study was not designed to make direct comparisons between paraprofessionals and nurses."1
Olds and colleagues attempted to statistically control post hoc for some of the observed nonequivalencies across nurse and paraprofessional home visitor groups occurring during the study’s implementation, most notably in the substantially greater turnover that occurred among the paraprofessional home visitors, as well as in a lower intervention dosage paraprofessional home-visited families actually received. However, as reported, a lack of equivalence across the paraprofessional and nurse home visitor groups remains across several additional important factors, preventing direct comparisons of nurses versus paraprofessionals without additional, more careful study. Several examples stand out:
First, paraprofessional home visitors were substantially younger (mean age = 33) than nurse home visitors (mean age = 41),2 and earlier research has indicated that worker age and previous work experience are key predictors of service effectiveness assessments.3,4,5
Second, nurses and paraprofessionals received nonequivalent training: "Because nurses were expected to exercise more independent judgment in helping mothers deal with physical health concerns, the nurses were given more in-depth training on the physical health and development of the mother and child."6 In addition, nurses were provided specific training in "solution-focused" therapy techniques, whereas paraprofessionals were trained in an "alternate problem-solving method."7
Third, paraprofessionals and nurses implemented different program protocols. Some of these differences were planned,1 while others were unplanned, resulting from paraprofessionals’ discomfort with implementing a protocol that was originally designed for nurses.7 Such changes resulted in significantly different emphases in the home visit contents delivered across paraprofessional and nurse groupings, with nurses providing greater attention to personal health during pregnancy and parenting during infancy,7 areas in which outcome differences were found.
Fourth, "a sense of competition emerged between the nurse visitors and paraprofessional home visitors," where "performance anxiety emerged among the paraprofessionals themselves that paralleled the anxiety they felt in the presence of the nurse home visitors."2 Such competition between treatment groups is a common threat to the internal validity, highlighted in Cook and Campbell’s classic work on quasi-experimental research as "demoralization in groups receiving less desirable treatments."8 Given the lack of equivalence on a number of important factors across nurse and paraprofessional home visitor groups, differences reported in the magnitude of effects are difficult to attribute solely to the professional status of the service deliverers.
Rather than noting these as study limitations, Olds and colleagues suggest that problems in the implementation of their study design were "inherent in paraprofessional programs" (page 493). They proceed to add: "Although other paraprofessional program models might perform better than the one tested here, the absence of clinically or statistically significant effects for most paraprofessional models tested in randomized trials makes this unlikely" (page 493). Olds and colleagues here do not acknowledge the previous studies of paraprofessional home visitors that have reported important intervention effects,9–11 as well as previous studies employing nurses as home visitors that have failed to report significant program effects,12,13 leaving the impression that the scientific base has been selectively attended to. A recent meta-analysis has examined home visitation trials targeting parent-child outcomes related to child maltreatment, and has reported that when considering the full array of studies employing nurses and paraprofessionals, program engagement and retention rates are virtually indistinguishable across professional status types, as are observed effect sizes.14
Most limiting for the broader field, the paraprofessionals used in the Denver trial do not appear as representative, holding substantially less academic preparation and relevant prior training than home visitors employed in numerous other home visiting programs operating across the United States. As noted about the Denver trial, "in order to highlight the contrast between paraprofessional and nurse home visitors ... [we] refrained from hiring [paraprofessional] applicants who possessed bachelors’ level education" (page 80).2 Paraprofessionals in the Denver trial held no academic credentials in relevant fields such as nursing, education, psychology, or social work, in direct contrast to paraprofessionals employed in some of the most widely disseminated home visiting programs in the United States, such as those within the Healthy Families America initiative15 or those employed in PAT ("Parents as Teachers") programs.16,17 A national study of Healthy Families America home visitors, for example, noted that 81% of the paraprofessionals held bachelor’s degrees or some college experience, and 10% had post-bachelor’s graduate training. As well, 85% had previous work experience in home visitation programs, most in the field of early childhood and child abuse and neglect, and 75% of the home visitors held specialized educational training in child development, social work, nursing, or education.18
Looking beyond the generalizability concerns of the Olds et al study, a number of other well-controlled home visitation studies have reported that nurses, paraprofessionals, and even graduate students can deliver home visiting services that provide positive benefit for at-risk families.14 Like any study, the Olds Denver trial taken forthrightly must be seen for its contributions as well as its limitations, and placed in the context of the full array of previous rigorously conducted home visitation studies. Taken as a whole, and not attended to selectively, the empirical evidence does not yet clearly indicate that one specific professional status is optimal over others in the delivery of effective preventive home visiting services.
REFERENCES
In Reply.—
Guterman, Anisfeld, and McCord urge readers not to overinterpret the findings of our trial, raise a number of issues regarding the equivalence of the nurses and paraprofessionals, express doubt that challenges with paraprofessional home visitation are as common as we have claimed, and point out that having nurses serve as home visitors does not ensure success. In sorting out our differences, it will be important for readers to note that our group conducted the Denver trial because most paraprofessional home visiting programs tested in randomized trials had produced sporadic, small effects that rarely were clinically or statistically significant.1,2 The study was designed to determine whether paraprofessionals’ performance could be improved if they were given excellent supervision and structured guidelines shown to work in previous trials when delivered by nurses.
We focused the Denver trial on a segment of the paraprofessional population that shared many of the social characteristics of the families they served, given the theory that reduced social distance between visitors and families would increase program effectiveness.3,4 We included the nurse arm in the trial to help us interpret the success or failure of the paraprofessional program in light of the nurses’ performance and to provide a third test of the Nurse Family Partnership (NFP), as the nurse program is now called.
Guterman et al are concerned about the nonequivalence of the nurses and paraprofessionals. Nonequivalence in the background of the families would constitute bias that challenges the results of the trial (there is none of any consequence in the Denver study). Nonequivalence in the background of the visitors is to be expected in light of the study’s purpose. We explicitly chose paraprofessionals whose backgrounds were "closer" to the backgrounds of the families they served, not to the backgrounds of the nurses. Compared with nurses, paraprofessionals thus had different educational and socioeconomic backgrounds, familiarity with the communities they served, and ages. Moreover, all of the paraprofessionals were parents compared with 70% of the nurses.5
Our primary objective in hiring paraprofessionals was to recruit the best paraprofessional visitors with these backgrounds and to provide them with excellent resources to serve their families. This included providing them with twice the level of supervision as the nurses, detailed visit-by-visit guidelines, and a different method of promoting parents’ abilities to cope with the demands of becoming parents and living in poverty. The paraprofessionals’ supervisors decided that the paraprofessionals would function best if they used the "problem-solving" method (a strategy used by nurses in our Memphis trial and that was better suited to the paraprofessionals’ lack of formal training in the helping professions) rather than being required to learn the "solution-focused" strategies used by nurses in Denver. Moreover, although the paraprofessionals spent less time learning about the physical health aspects of the curriculum, they were taught to encourage the parents to work with their primary care providers in addressing physical health issues.5
These relatively minor, planned differences in the content and methods of working with families were necessary, given the backgrounds of the visitors. Aside from these planned differences, deviations that emerged in the conduct of the program resulted from the unique ways the 2 visitor types used the guidelines, given that visitors and families were free to modify the frequency of visitation and content of the program as needed. As we (Hiatt et al) have noted, the paraprofessional visitors did not feel comfortable in using the parenting portion of the curriculum because it felt "foreign and unnecessary" to them. Part of their discomfort was attributable to their being expected to "teach" parents how to care for their children, an activity some felt was patronizing.5 Teaching is a natural part of nurses’ roles. This probably accounts for the lower portion of time paraprofessionals, compared with nurses, spent helping parents learn how to care for their children.6
Analyses of time spent on various program content areas, including parenting, physical health, and environmental health, however, showed that after controlling for other family background characteristics, amount of time spent on specific activities did not account for differences in outcomes between nurse-visited and paraprofessional-visited families; indeed, an increased amount of time spent on helping parents learn how to care for their children was associated with poorer language development in the paraprofessional group—probably a reflection of the paraprofessionals’ spending more time on this topic with families who especially needed this kind of help.7 Their underemphasis of parenting in conducting the program, by itself, does not account for the paraprofessionals’ negligible effects on child outcomes.
It is interesting that Guterman, Anisfeld, and McCord have failed to acknowledge the major thrust of the article in which we note the paraprofessionals’ performance anxiety: that despite their commitment to and identification with the families they served, a significant portion of the paraprofessional visitors exhibited problems in maintaining appropriate boundaries in working with families, in working effectively with one another, and in making good use of the supervision given to them.5 These factors, we think, are more likely to have played a role in explaining the paraprofessionals’ underperformance than is their anxiety, which is equally likely to have improved their effectiveness.
Problems with boundaries, working with others in professional ways, and making good use of supervision are common in paraprofessional programs when visitors share many of the characteristics of the families they serve. The literature supports this conclusion,8–10 as does our experience in consulting with others who have tested paraprofessional programs (R. Maynard, personal communication, November 2002). Moreover, we believe that nurses have more persuasive power compared with other visitor types during pregnancy and infancy because of their legitimacy and value in the eyes of pregnant women and new parents, who are particularly concerned about physical health issues.
We are thoroughly familiar with the 2 trials Guterman, Anisfeld, and McCord cite as evidence that paraprofessional programs can work. One consists of a trial of a single gifted home visitor who grew up in the Baltimore neighborhood in which she worked, but by the time she served as a home visitor, held a bachelor’s degree.11 Is the success observed in this trial a reflection of paraprofessional programs overall or the unique talents of a single individual? The second trial, conducted in Denver, experienced problems with implementation of the design (treatment-based differences in rates of attrition and a control group that was not entirely randomized) and produced effects of questionable clinical importance (no effects on prenatal health, maternal life-course, and child maltreatment, but improvements in observed interaction between mothers and their children—effects limited to dyads in which the mothers were teens and Hispanics).12–14
We have noted in our reviews of home visiting programs that simply hiring nurses to serve as home visitors is insufficient1,2 and have referenced the very studies cited by Guterman, Anisfeld, and McCord to support this position. When the outcomes of interest are prenatal health, child maltreatment, childhood injuries, and maternal life course, programs that have used nurses have produced the most dramatic and consistent effects, but only when they follow an effective program model. The NFP tested in the Elmira, Memphis, and Denver trials has produced the largest and most consistent effects on these outcomes of any home visiting program examined to date.15,16
Part of the NFP program model requires that the nurses serve only low-income first-time mothers (and their families)—as they are in need and more likely to be receptive to such services than are higher income and multiparous women. The current trial supports our position that you need at least 2 (and probably 3) components to produce the greatest effects on these outcome domains: 1) an effective program model; 2) the right visitor type; and 3) a target population that is in need and sees the value of the service being offered.
It is difficult, using the kind of meta-analyses conducted by Guterman,17 to discern requirements for program success because they consist of combinations of program characteristics. Unless there are large numbers of studies to amply fill the cells in a cross-classification matrix of program models, visitor backgrounds, and population characteristics, such analyses will fail to discriminate the conditions necessary for success. Meta-analysis, in the context of limited samples of studies, is a blunt instrument.
Guterman, Anisfeld, and McCord are right that today many home visiting programs, such the Parents as Teachers and Healthy Families America programs, have moved toward recruiting visitors with higher educational qualifications. When put to careful tests in randomized, controlled trials, however, these program models have produced few clinically or statistically significant effects.18–20
Results of other recently reported trials of paraprofessional programs corroborate the small effects detected for the paraprofessionals in Denver. The Early Head Start programs that consisted of home visiting alone produced effects that also were small (typically in the 10%–15% of a standard deviation range at best).21 This work is important because the Head Start visitors were held to very high standards of program implementation, usually had higher educational backgrounds, had caseloads roughly half the size, and were required to visit on a more frequent basis compared with paraprofessionals in the Denver trial.22 Similarly, a well-conducted trial of an augmented version of the Healthy Families America program recently has been completed in San Diego; in this trial, parenting classes and support groups were added to the home visiting intervention, the visitors were required to have a minimum of an associate’s degree and 4 years of previous work in health and human services, and they were able to retain families in the program at higher rates than most other home visiting programs that use paraprofessionals as home visitors.20 This program produced effects similar to those achieved by paraprofessionals in Denver for comparable outcomes.20
We hope that successful paraprofessional program models will be developed, tested, and replicated—especially for segments of the population that have special needs and for whom the evidence is less clear. The Best Beginnings Plus program for substance-abusing parents, tested by Dr Anisfeld,16 and another paraprofessional program for substance-abusing parents in Seattle,23 for example, may eventually meet these standards. In the meantime, vulnerable families deserve services that have the best chance of helping them and taxpayers deserve investments of public funds in prevention programs proven to work.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2003 American Academy of Pediatrics. All rights reserved. |
||
|
||
