Paraprofessionals' efficacy shouldn't be taken on trust

February 1, 2003

Dr. David Olds' intensively researched nurse-visitor program for mentoring first-time mothers is one of the few demonstrated cost-effective success stories in social policy. Nurses begin visiting the women during pregnancy, the earlier the better, and continue regular visits until the children are 2 years old.

The results have been sufficiently impressive that Colorado has implemented the Nurse-Family Partnership with money from the tobacco settlement, though the state's budget troubles may limit the scope of the program in the next budget. That would be a pity.

The program has been tested in randomized, controlled trials, with one in Elmira, N.Y., starting as long ago as 1978. Those children have been followed through adolescence, and those whose mothers had nurse visitors were significantly less likely to develop problems than those in the control group. A second trial in Memphis, Tenn., refined the program, and the most recent trial, in Denver, specifically tested whether using paraprofessionals trained in the program as visitors worked as well as the original model where the visitors were registered nurses.

It didn't, and the result has enormous implications for other social programs that employ paraprofessionals under the assumption that they will relate better to the population targeted by the program. In too many cases, that assumption has never been tested.

Olds is on the faculty of the University of Colorado Health Sciences Center. He and several co-authors published their results in the September issue of the journal Pediatrics. From March 1994 through June 1995, the Denver trial of the Olds program recruited approximately 750 low-income women who were assigned either to a control group, a group that received visits from paraprofessionals or a group that received visits from nurses.

In the paraprofessional group, the only statistically significant result was that mothers with "low psychological resources" interacted better with their children than their counterparts in the control group.

In the nurse group, as in previous trials, there were many statistically significant results, including some that have a bearing on the children's subsequent history. Mothers smoked less, worked more and were more likely to delay having a second child. Their children were less likely to be delayed in language development, and ranked higher on measurements of emotional and mental development.

The cost of the program, per family, was roughly $6,000 for those in the paraprofessional group, and $9,000 for those in the nurse group. But if there are no long-term benefits to receiving visits from paraprofessionals, then the paraprofessional money is entirely wasted, while the larger amount spent on nurse visits has been shown, in a RAND Corp. study, to return about $4 in benefits for each $1 spent.

The study was not intended to prove that nurses are better than paraprofessionals; instead, it was intended to find out whether paraprofessionals could improve their performance "if given structured guidelines, excellent training and supportive supervision." It does not disparage or demean their work. It shows their efforts had a smaller effect, but not why. They could have been just as dedicated and caring as the nurses, and as well-trained in the specifics of the visitor program, and yet less effective because the mothers responded better to the aura of medical authority conveyed by the nurses' credentials.

The question is, what about other programs that use paraprofessionals? Not only other home-visit programs, but, for example, Head Start.

Head Start is a vastly expensive sacred cow, and yet in studies spanning more than a quarter-century, it has no demonstrated long-term educational benefits. Could that be because, in many cases, it functions more as a neighborhood jobs program than as an educational experience for children? Shouldn't somebody try to find out? Rather than pushing to enroll more children in a program unlikely to do them any lasting good, we might see more results if there were fewer children in Head Start, but the ones who were enrolled were actually educated by people qualified to do it.

Another popular use of paraprofessionals is as teacher aides. Teachers very much like having aides, and the aides themselves presumably like having jobs. And by all evidence, they are as devoted to the children in their classrooms as the teachers are. But what are they accomplishing? A large-scale study in Tennessee tested smaller classes against larger classes with and without teacher aides. The small classes made a difference. The aides didn't. Where should states put their educational resources?

"Effective paraprofessional models may eventually be developed," the study concludes, but their limited effects "sound a cautionary note."

Indeed. Is anybody listening?