3.3. ESTUDIO GEOTÉCNICO
3.3.2. ENSAYOS DE LABORATORIO
3.3.3.10. Valor Relativo de Soporte (CBR)
The question of follow-up is another contentious issue^®. Many financially- constrained projects leave out follow-up assessment, but this may be a false economy because we then have no way of knowing whether treatment effects can be maintained over time and, therefore, whether the treatment is valid in its own right or whether transient effects were merely transmitted through, perhaps, the charisma of the therapist or other 'non-specific effects'
(Frank 1965). Some reviewers (Casey and Berman 1985; Nicholson and Berman 1983; Frank 1972) have found results at follow-up to be so highly correlated w ith assessment at termination of treatment (p < .001, as analysed by Nicholson and Berman) that they maintain that follow-up is an
unnecessary addition to any research s tu d y ,a d d in g nothing new to the
’®Follow-up has additional complications for psychoanalysis because of the potential dangers of plugging into, and thus perpetuating, transference fantasies.
^°But, as Wallerstein points out, 'Adequate follow-up study is recognised everywhere in clinical medicine as a most vital aspect of the understanding of the course of an illness and the success of its treatment. In cancer research, the five-year follow-up has been widely institutionalised; before this point one rarely speaks, even cautiously, in terms of probable cure' (Wallerstein 1986). Studies that reject the necessity to follow-up may be leaving out an essential component of all competent treatment evaluation.
information gained at termination. Although the average length of time until follow-up in the studies reviewed by Nicholson and Berman was 872 months, which is arguably not long enough for differences to emerge, the range varied between 1 and 114 months and no differences were found that might have been dependent on the time scale.
Other work has reached different conclusions, advancing evidence which suggests that evaluation of effect may change considerably, in the direction of either increased or reduced improvement, over time. In 1981, Kolvin and his colleagues published the results of a school-based treatment for neurotic and conduct disorders. The measures at the end of treatment (which
included a number of different therapies) showed, inter aiia, fe w
improvements for either the neurotic or the anti-social children w ho were treated w ith group therapy or behaviour therapy. However, 18 months later, these groups showed marked improvement. The initial conclusions that the treatment had had no impact needed to be revised. Some meta-analytic work, too, has found a greater therapeutic effect size 5-6 months after the ending of treatment (0.93 compared w ith 0.79). This supports the
observation that patients leaving therapy 'uncertain and discontented w ith its results may realise benefits many months or even years after therapy has ended' (Sargent in Nicholson and Berman 1983).
The reverse may equally be true and Kazdin's review shows that it is
sometimes the case that 'the treatment that appeared more or most effective at post-treatment did not retain this status at follow -up'. Goldstein et al (1979) and Smith, Glass and Miller (1980) found that there were many more studies reporting positive results at the end of therapy than there were
reporting positive results at follow-up.
Follow-up assessments in any outcome evaluation of child therapy are particularly indispensable because of maturation effects in children. It is possible that 'treatments that appear effective or differentially effective in the short run may not surpass the impact of developmental changes' (Kazdin
Most of the therapies examined in the above review papers are behavioural in approach. Psychoanalysis requires different criteria to be evaluated in its exploration of maximum effectiveness. In place of symptom alleviation, analysts look for the 'accomplishment of an essential personality change which is the outcome of a redistribution of psychic energy' (Smirnoff 1971). None of this is well-defined or easily operationalisable but the importance of fundamental, integral changes in analysis is clearly implied. Therapies that make a good showing at immediate post-treatment assessment but whose effects decline rapidly from that point might, we could speculate, have offered greater possibility of permanent modification of pathogenic cognitions, affect or behaviour had they been conducted over a more
extended period of time. This offers important possibilities for treatments like psychoanalysis in which change is asserted to take effect only over long time scales.
The little available research evidence suggests that psychoanalytic outcome does indeed vary over time, at least w ith adult patients. Thomas
McGlashan's retrospective study of 446 adult patients and their psycho analytic treatment at Chestnut Lodge, Maryland, demonstrates the importance of follow-up at various points after treatment, ideally over a period of some length (McGlashan 1984a;1984b;1986a;1986b;1986c). It was already known that treatment effects can extend over a period of at least tw o years after the end of an active phase of therapy (Kolvin et al 1981) but McGlashan found significant changes in functioning in certain diagnostic groups over much longer time-scales. General indications that the majority of patients w ith affective disorders had tended to improve by follow- up and the majority of schizophrenic sufferers to deteriorate are not
discordant w ith epidemiological data. The outcome of these patient cohorts varied little over time. That is to say, the depressed patients maintained improvement while the psychotic patients stabilised somewhat although the trend remained downward. The unexpected finding was the long-term
outcome results from patients diagnosed as fulfilling the criteria for Borderline Personality Disorder. For this group, global functioning does seem to vary over time on a shallow inverted U-curve. Borderline patients are at their best
in the second decade after discharge from hospital and treatment
(McGlashan 1986a). This may well be attributable to age or other variables and have nothing to do w ith the treatment itself but it does underline the importance of looking for cohort effects and of attempting to determine which diagnostic groups at which age benefit most from which treatment.
Wallerstein, whose outcome study of adult psychoanalysis has already been cited, also found changes in functioning between termination and the very intensive and extensive follow-ups which continued for up to 30 years (Wallerstein 1986)! Eighteen out of his 42 patients consolidated and
enhanced their position at termination; most of these had been assessed at termination as having made significant improvement, but this was not so for all the patients and one had been categorised as a failure at the end of treatment. Ten demonstrated no change at follow-up and 10 regressed, including tw o whose achievement at the end of therapy was classified as 'very good'. (Of the other 4, 1 had died and 3 had become 'therapeutic lifers', still in treatment of one sort or another.) It is obvious that although the future of many patients can be accurately predicted at termination this is by no means true for all. And the unexpected operates in either direction.
3 .4 .V Diagnostic profiles
One of the essentials in treatment outcome methodology is precise
specification of the diagnostic categories into which patients fall. Different symptom profiles may respond in totally different ways to a particular intervention. The patients in some of the earliest reviews (eg Levitt
19 5 7 ;1 963) were broadly classified as suffering from 'neurosis'. This goes some way to providing useful information because it is known that 'neurotic' children respond better to all types of treatment than do delinquent or
psychotic children (Levitt 1963) and we might therefore be justified in asking whether the results obtained could be regarded as biased in favour of
treatment.
This is specially important to know if the conclusions are illegitimately generalised to the whole population of referred patients, because the most
common reason for referral to a child guidance clinic is conduct disorder and other aggressive and disruptive behaviours (Task Force on Intervention Research 1988) which cannot adequately be subsumed under a 'neurotic' label. But a term like 'neurosis', though excluding some behaviours, is still not nearly rigorous enough to be used as an operational research concept. It would normally be regarded as including all 'internalised' symptoms such as anxieties, depressions, phobias, withdrawn and obsessional behaviour - a heterogeneous collection of problems indeed. It is as meaningless to measure effectiveness of psychotherapies by combining a variety of diagnoses as it is to evaluate general medicine by assuming homogeneity between diseases as dissimilar as diabetes, cancer and pneumonia (Kazdin 1988). Chronic illness, viral infections, immune deficiency syndromes, all require different
treatments and all have different prognoses. So it is w ith psychological disorders.
Diagnostic criteria have become more strictly defined and standardised over the past 15 or 20 years so that it is now no longer necessary (nor would it be methodologically acceptable) to use vague descriptions such as
'emotionally disturbed' or 'disruptive' to refer to subject-patients. In the 1960's attention was drawn to the vast differences in diagnostic patterns between countries, regions, hospitals and individual clinicians (e.g. Lader 1977) and it became imperative to produce specifications of particular
syndromes that were universally recognisable. Consequently, revisions of the Diagnostic and Statistical Manual for psychiatric disorders have become increasingly explicit in category specification. The first edition in 1952 identifies just over 100 categories into which symptom profiles might fall. Sixteen years later the classifications had increased to more than 180. The most recent revision of the latest edition (DSM-III-R 1987) offers more than 260 possible psychiatric diagnoses w ith strict criteria to be fulfilled before a particular diagnostic label can be applied.
There is now much less excuse for vague definition of patient category than there was in the past; however, assigning patients to the correct diagnostic category for research purposes is still very much a live issue. There is
increasing evidence that children who meet the criteria for one disorder are very likely to meet criteria for other diagnostic categories as well (Kazdin
1990). This 'co-m orbidity' presents problems for treatment evaluation. The principal diagnosis will probably be selected as the focus of treatment but an apparently homogenous group of children may respond to their treatment according to secondary diagnoses such as depression or hyperactivity. Unless all possible factors are taken into account the results may be
impossible to interpret correctly. There may be important indications for and against applications which have been overlooked by conflating tw o groups of patients who seemed, on the surface, similar.
There are no 'easy' solutions to such problems; individual factors, many of which fall outside the reach of diagnostic and clinical scrutiny and may pass unrecognised, mediate treatment outcome for all patients in some way or other. Recent meta-analyses of child psychological treatments (Casey and Berman 1985; Weisz et al 1987) indicate the depth of the complexity w ith suggestions that effective outcome is dependent on a variety of patient, therapist, and treatment characteristics. We are making slow progress towards an understanding of how best to measure and assess psychological treatment outcomes, and how to combine methodological exactitude with clinical sensitivity and non-violation of treatment integrity, but we are still a long way from being able to give any sort of definitive answer to the
question, 'W hat treatment, by whom, is most effective for this individual w ith that specific problem under which set of circumstances?' (Paul 1967).
3 .4 .vi Maturational effects
Maturational effects in childhood raise at least tw o important issues for outcome research. Firstly, diagnosis of clinical problem cannot be made independently of the age of the child. Problem behaviours are highly
correlated w ith age (Schwartz and Johnson 1981), younger children showing greater levels of instability because of their uncertainty about, and
dependence on, the outside world. Some behaviours that might be generally considered as disturbed or maladaptive are so commonplace in young
teenager w ill all require different assessment and treatment. The practice of combining children of all ages into evaluative reviews of treatment efficacy has little value.
Secondly, assessment of therapy outcome has to take into account maturation of the child during the time of treatment. Many childhood
problems do alleviate w ith time w ith or w ithout treatment (MacFarlane et al 1954). Differentiating between problems which will disappear of their own accord and those which require intervention to prevent consolidation and permanent adoption as part of a personality structure is extremely difficult. In the few evaluations of adult psychoanalysis that have been published, it proved impossible to predict at the beginning of treatment which patients would benefit and which would not (Wallerstein 1986.). How much more difficult, then, w ith children for whom constant change and development is part of the natural maturational process. The longer the treatment the more difficult it becomes to separate maturational from therapeutic effects.