To determine the effectiveness of interventions targeted at providers to enhance the use of mammography.

English language studies were sought in MEDLINE from 1980 to 1998. Details of the search strategy were given. Reference lists from selected studies were scanned. Handsearches were conducted of the following: Journal of General Internal Medicine; Medical Care; Preventative Medicine; Annals of Internal Medicine; Archives of Internal Medicine; Cancer Epidemiology; Biomarkers and Prevention for June to August 1998.

Study designs of evaluations included in the review

Studies in the United States that were prospective follow up studies with either randomised assignment to an intervention or control group or a non randomly selected concurrent control group were included. Pre/post designs without controls and uncontrolled trials were excluded as were abstracts.

Specific interventions included in the review

Inclusion criteria stated that cognitive, behavioural and sociological interventions designed to increase mammography use were to be included. Cognitive interventions were considered to include the provision of new information and education, increase existing knowledge, and clarify misconception. Specific interventions included audit with feedback and educational sessions or materials. Behavioural interventions altered cues or stimuli associated with screening behaviour and included reminders or administration office systems. Specific interventions included a reminder or office system prompts.

Sociological interventions used social norms or peers to increase screening adherence and included audit. Specific interventions included nurse based interventions or reorganisation of the clinic. Interventions that altered the structure of care delivery were classified as sociological.

Interventions with multiple strategies were classified as behavioural and cognitive.

Control groups were defined as active controls which provided a lower level of intervention and usual care controls in which no intervention associated with mammography was performed. Interventions were targeted at providers, providers and patients or communities. Studies were conducted in University or Community settings.

Participants included in the review

Providers of mammography screening were included. The majority of study populations were white women aged 50 years or over with rates of previous mammography, where stated, ranging from 25% to 49%.

Outcomes assessed in the review

Rates of ordering screening and completion rates of screening were assessed using patient self-reports, medical charts, completed mammography reports from radiology report, and ordered/response to reminder.

How were decisions on the relevance of primary studies made?

The authors do not state how the papers were selected for the review, or how many of the reviewers performed the selection.

The authors do not state that they assessed validity.

The following data were extracted onto a standardised data extraction form: type of intervention; characteristics of mammography outcome determination; the patient population; and intervention effectiveness. The effect size and 95% confidence intervals (Cis) were calculated. For RCTs effect size was calculated as the difference in mammography rates between the intervention and control groups. For non-randomised concurrently controlled trials the effect size was calculated as the difference between rates post-intervention and pre-intervention for the intervention group and the control.

The authors do not state how data were extracted for the review, or how many of the reviewers performed the data extraction.

How were the studies combined?

The DerSimonian and Laird random-effects summary statistic and 95% confidence intervals were calculated separately for difference types of intervention.

How were differences between studies investigated?

The DerSimonian and Laird Q statistics were used to assess homogeneity for interventions grouped by mechanism of action and the type of control group. The influence of a single intervention on summary results was explored by sequentially dropping each intervention and recalculating the summary statistic.

Thirty-five studies including 23 randomised controlled trials, (RCTs) were included.

Overall there were 23 behavioural interventions, 8 cognitive interventions, 5 sociological interventions, and 13 interventions that combined behavioural and cognitive approaches. Most studies were RCTs in university settings (23 studies).

Behavioural interventions (included a reminder or office system prompts):

Provider targeted with usual care control (7 studies with 3559 participants including 6 RCTs): interventions increased mammography. Q = 16.5. Effect size = 13.2 (95% CI: 7.8, 18.4).

Provider targeted with active control (7 studies with 25545 participants including 6 RCTs with 9811 participants): interventions increased mammography. Q = 17.4. Effect size = 6.8 (95% CI: 4.8, 8.7).

Provider and patient interventions with usual care controls (2 studies with 213 participants): interventions increased mammography. Q = 0.0. Effect size = 20.5 (95% CI: 9.7, 31.3).

Provider and patient interventions with active controls (4 RCT with 6434 participants): interventions increased mammography. Q = 17.3. Effect size = 8.9 (95% CI: 3.1, 14.6). Cognitive interventions:

Provider targeted with usual care control (4 studies with at least 2013 participants, including 2 RCTs): interventions increased mammography. Q = 4.3. Effect size = 18.6 (95% CI: 12.8, 24.4).

Provider and patient interventions with usual care controls (1 RCT with 432 participants): interventions increased mammography. Effect size = 16 (95% CI: 7.3, 24.7).

Community targeted with active controls (3 studies with at least 4437 participants, including 1 RCT): interventions increased mammography. Q = 7.8. Effect size = 9.6 (95% CI: 3.4, 15.8).

Sociological interventions:

Provider targeted with usual care control (2 studies with 4932 participants): inconsistent results.

Provider targeted with active control (3 studies with at least 1208 participants, including 1 RCT): interventions increased mammography. Q = 3.0. Effect size = 13.1 (95% CI: 6.8, 19.3).

Interventions with combined modalities:

Provider targeted with usual care control (4 studies with at least 1198 participants): interventions increased mammography. Q = 9.8. Effect size = 21 (95% CI: 8.8, 33.6).

Provider and patient targeted interventions with usual care controls (4 studies with 2994 participants): inconsistent results.

Provider and patient targeted interventions with active controls (2 studies with 2392 participants): interventions increased mammography. Q = 9.1. Effect size = 16.1 (95% CI: 11.6, 20.7).

Community targeted with usual care controls (3 studies with at least 4437 participants): interventions were not associated with increased mammography. Q = 7.7. Effect size = 1.1 (95% CI: -6.8, -9.0).

Interventions targeting both patients and providers were not significantly better at increasing screening than those targeting providers alone, and multiple approaches were generally not more effective than a single approach. All interventions targeted at physicians were effective in increasing screening rates. Decisions to use a particular approach will depend on resources, expertise, feasibility, and cost effectiveness.

The aims and inclusion criteria were stated. The search included several databases. Statistical homogeneity was assessed and Q-statistics presented though without p-values. Some sensitivity analysis was undertaken. The discussion included consideration of the following limitations of the review: combining of data from studies conducted in different populations and environments; inconsistencies in the unit of analysis; multiple interventions from the same study; inability to measure the penetrance of the intervention; potential patient and provider selection bias; publication bias; and lack of data on intervention durability.

Limiting included studies to those published in the English language may have omitted some relevant studies. Methods used to select studies and extract data were not described. Insufficient detail was given of individual studies to allow an assessment of the clinical homogeneity of actual interventions, settings, populations and outcomes measures used. Validity was not assessed. Studies were grouped and analysed according to the nature of the intervention and control groups without consideration of the study design and study quality. Studies reviewed were limited to those conducted in the United States and it is uncertain if these results would apply to other countries with different health care systems.

In view of the lack of assessment of the validity of the studies on which the evidence was based, and the limitations as discussed by the authors, these results should be interpreted with caution.

Practice: The authors state that decisions to use a particular approach will depend on resources, expertise, feasibility and cost-effectiveness.

Research: The authors state that research is required into the effectiveness of different interventions in patient subpopulations such as the elderly, minorities or those on low income, and the costs of providing these interventions.

Agency for Health Care Policy and Research, grant number HS08395; Department of the Army, grant number DAMD17-94-1-4212; Breast Cancer Center grant; National Cancer Institute, grant numbers RO1 CA72908 and 2T32 CA09314.

1. Mandelblatt JS, Kanatsky PA. Effectiveness of interventions to increase enhance physician screening for breast cancer. J Fam Pract 1995;40:162-70. 2. Yabroff KR, Mandelblatt JS. Interventions targeted to patients to increase mammography use. Cancer Epidemiol Biomark Prev 1999;8:749-57.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.