Mitch Medical Healthcare

Limitations in the Use of Meta Analytic Techniques

Attempts have been made to gauge the effect sizes of specific intervention strategies by combining studies, using meta-analytic methods. However, assessment methods for nonadherence vary across studies, and the efficacy of a specific treatment is dependent on the way the outcome measure (adherence) is defined. For example, in a recent multicenter study

Table 13-2. Summary of approaches used to enhance treatment adherence

Treatment modality

Indications

Individual responsible for implementation

Preventive educational approachesa

Family-based approaches Behavioral treatment

Specific psychiatric or psychological interventions

Changes in medication regimen

Changes in organization of structures beyond the patient and familyc

All patients who are seen in a clinic

Patients whose families are at risk (e.g.. history of abuse) or are known to be distressed or dysfunctional

Patients who are known to be nonadherent

Patients who are suffering from a known psychiatric disorder or who have behavioral health risks

Patients whose nonadherence is thought to be related to either the side-effect profile of a medication or difficult administration

Patients whose nonadherence is related to factors such as clinic settings or insurance status

Clinic staff

Care provider familiar with family systems approaches to treatmentb

Clinic staff and family members, with consultation with a mental health provider

Psychologist, psychiatrist, or clinical social worker

Medical team

Social workers and clinic staff

Nonspecific.

bA skilled individual is seldom available in clinical settings to administer these treatments. cAn example: helping the patient to get medication coverage benefits.

(Kato et al. 2008), 375 pediatric oncology patients were randomly assigned to receive either a video-based intervention to improve adherence or a video game (without the intervention) alone. In this study, adherence was assessed by four different methods. Results from the self-report tool, which was administered to all of the patients, determined that the intervention did not improve adherence. However, the intervention seemed to have worked for a small subgroup that used electronic monitors. In an even smaller group, in which medication metabolites were reviewed as indicators of adherence, one metabolite (6-mercaptopurine) but not the other (6-thiogua-nine) showed a significant improvement. Furthermore, no significant correlation existed between those four adherence measurement methods.

If the same study finds strikingly different treatment results when different methods of adherence detection are used, certainly grouping together different studies in a meta-analysis is risky and might result in estimates that are inaccurate. Thus, when the efficacy of a specific treatment strategy is considered, it is important to know what method was used to measure adherence. Meta-analyses, therefore, should be used very cautiously, if at all, when evaluating adherence research. They can provide support for specific interventions only if the meta-analyzed studies are comparable (i.e., replication studies), and they can serve as important tools to highlight general principles that can form the basis for future studies. However, they should not be interpreted as conclusive evidence for treatment efficacy and should not be used to try to determine whether a certain treatment is more promising than others.