Table 1.
Details of included studies
| Reference | Country | HCP | Type of study | Cancer | Aim of intervention | Details of intervention | Outcome of study | Problems with study | QA score achieved by consensus |
|---|---|---|---|---|---|---|---|---|---|
| Argenziano et al, 200613 | Italy and Spain | Primary care physicians (PCPs) | RCT | Skin | To determine whether PCPs achieve greater accuracy when triaging skin lesions suggestive of skin cancer using dermoscopic evaluation and the 3-point checklist in addition to the standard clinical examination | Intervention components: the study was divided into four steps; a 1-day training course described the ABCD rule for clinical diagnosis and basic criteria for non-melanoma skin cancers and described the 3-point checklist (simple diagnostic algorithm); in step 2, participants were randomly assigned to assess lesions either by standard clinical examination or by a a hand-held dermatoscope, and consecutive patients asking for screening or exhibiting skin tumours were seen by the PCPs for routine examination; step 3 involved re-evaluation of the patient by dermatologists; and step 4 involved definitive diagnosis by dermatologists. Number of participants recruited: 88. Number of participants completing study: 73. Intervention delivered by: not stated. Length of intervention: 1 day. Time to follow-up: M/A. How intervention was assessed: patients assessed by the participants were re-evaluated by at least two melanoma experts—lesions lesions were evaluated and scored as benign or suggestive of skin cancer; all lesions suggestive of skin cancer were then excised and diagnosed histopathologically to assess participants' referral accuracy | The use of dermoscopy allowed PCPs to perform 25% better triage than naked-eye examination alone, and increased referral sensitivity and accuracy; increased negative predictive value (NPV)—participants were less likely to miss skin cancers | Selection bias; no method of randomisation; no participant details; patient population involved screened as well as presenting patients | 5 stars |
| Bedlow et al, 200014 | England | GPs | Before-and-after study | Skin | To evaluate the use of an illustrated booklet supplemented by a didactic lecture on the diagnosis of benign, borderline, and malignant lesions by GPs | Intervention components: two sets of clinical slides showing 10 malignant, 10 benign, and 10 border line lesions were shown. A lecture on the clinical features of such lesions and a booklet that the lecture was based on was then given out following the lecture. Number of participants recruited: 23. Number of participants completing study: 17. Intervention delivered by: not stated. Length of intervention: not stated. Length of follow-up: 2 weeks. How intervention was assessed: GPs attending a seminar on skin cancer were assessed without priorwarning; 2 weeks later, at an unrelated seminar, another test of 30 slideswas shown | Significant improvement in diagnostic ability of GPs following the intervention; 81% of the GPs assessed read the booklet between tests | Selection bias; short follow-up; small sample size; no details on how test slides were assessed (multiple choice); difficult to gauge impact of booklet (only read by 81% of participants, no record of how often read, etc); no information on who presented the lecture | 0 stars |
| Brochez et al, 200115 | Belgium | GPs | Before-and-after study | Skin (malignant melanoma) | To see if GPs' clinical ability to diagnose increased after amelanoma lecture (secondary outcome) | Intervention components: lectures on risk groups, clinical aspects, differential diagnosis, and management were given. Number of participants recruited: 160. Number of participants completing study: 146. Intervention delivered by: not stated. Length of intervention: not stated. Length of follow-up: none (immediate post-test). How intervention was assessed: test consisted of slides of different lesions with a brief patient history. GPs had to choose from seven possible diagnoses | Diagnostic accuracy of GPs improved from 49% to 56% following the course; 63% believed their diagnostic ability had improved as a result of the course | Selection bias; no further follow-up; small sample size; no other information such aswho carried out the lectures, what format they were given in, or how many therewere | 5 stars |
| Carli et al, 200516 | Italy | GPs | Before-and-after study | Skin (malignant melanoma) | To evaluate the effect of a 4-hour training session on classification and management of lesions by family doctors | Intervention components: Training session consisted of instruction in basic melanoma triage, summary of melanoma and skin cancer epidemiology, prevention and counselling, presentation of images of lesions, and major clinical simulators. Number of participants recruited: 41. Number of participants completing study: 41. Intervention delivered by: two dermatologists experienced in melanoma screening. Length of intervention: 4 hours. Length of follow-up: none (immediate post-test). How intervention was assessed: the pre-test contained 15 images of lesions for which clinicians had to select the correct diagnosis, and the same test was administered after the intervention | Post-intervention, diagnostic accuracy increased from 46.8% to 76.2%. No change in referral rate but suggestion of more appropriate lesion selection | No follow-up; small sample size; selection bias | 5 stars |
| Cockburn et al, 200117 | Australia | GPs | Before-and-after study | Breast | Evaluate the impact of guidelines, audit, education, and feedback on investigations of new breast symptoms | Intervention components: participating doctors completed an audit of the management of all women attending with a new (incident) breast symptom over a 12-week period before and after the intervention. The intervention consisted of feedback from the first audit from their own practice and grouped data from other practices, and a seminar on the guideline recommendations and evidence. Number of participants recruited: 227. Number of participants completing study: 104. Intervention delivered by: breast specialists and study authors. Length of intervention: 5–7 months. Length of follow-up: approx 6 months (unclear). How intervention was assessed: via feedback fromthe audits | Guideline adherence improved following the intervention, but already appeared to be quite high at the first audit; statistically significant improvement in five criteria | Selection bias; No control group; no details about who presented the audit findings orwhat format this took; authors acknowledge that participating GPs had more of an interest in breast health and only 18% completed both audits (high dropout rate); not consistent with UK guidelines | |
| DeGannes et al, 200418 | Canada | Family physicians | RCT | Skin | To evaluate the effectiveness of a video intervention to promote early detection of skin cancer | Intervention components: participants were randomly assigned to the intervention or control group. The intervention group was given an educational video that participant could watch as of ten as they liked for 6 months. It contained a 12-minute video of skin-cancer, prevention and who is at risk. Nothing for control group. Number of participants recruited: 52. Number of participants completing intervention: 27. Intervention delivered by: video. Length of intervention: 12 minutes. Length of follow-up: 6 months. How intervention was assessed: both received the same internet-based multiple-choice questionnaire pre- and post-intervention, which contained knowledge, prevention, and case-management questions. A skin biopsy audit of all biopsies submitted by participating GP swas also conducted for 6 months pre- and 6 months post-intervention | No statistically significant improvement was observed between the intervention and control group | No follow-up; high dropout rate; the intervention group was already better at diagnosing lesions before the intervention; no record of how many times each participant watched the video - may have had an impact on outcome; the control group was sensitised to cancer through participating in the study | 8 stars |
| Dolan et al, 199719 | US | Primary care physicians (internal medicine house staff and attending physicians with outpatient practices) | RCT | Skin | To test whether providing a brief education programme could change skin cancer control attitudes, improve knowledge, and increase counselling and examination among performance patients at moderate to high risk | Intervention components: physicians were stratified by training level and randomised to either the intervention or control group. The intervention group consisted of educational sessions on skin cancer control. One session concerned determining individual risk and targeting prevention strategies, and the other early detection. No details of control group. Number of participants recruited: 96. Number of participants completing intervention: 25. Intervention delivered by: a dermatologistand general internist. Length of intervention: 2 × 1-hour seminars. Length of follow-up: 1 month. How intervention was assessed: both groups filled in a questionnaire before the intervention and 1 month after, which consisted of skin cancer attitudes, beliefs, knowledge, and practice, previous dermatology training, risk factors, and lesion identification | Intervention had no statistically significant effect on skin cancer control attitude, beliefs, knowledge, or behaviours | Self-selected sample; high dropout rate; control groupmay have been sensitised by participation; difficult to determine whether participants were fully trained PHPs | 7 stars |
| Gerbert et al 200020 | US | Primary care physicians | Before-and-after study | Skin | To determine whether decision-support software can help primary care physicians proficiently triage lesions that are suggestive of basal cell and squamous cell carcinoma | Intervention components: the software consisted of a clinical information form, decision tree, and support features. The algorithms for the decision tree incorporate the patient's clinical information. The test stimuliwere of 30 digitalised test images of cancerous, pre-cancerous, and benign lesions. Number of participants recruited: not stated. Number of participants completing intervention: 20. Intervention delivered by: developed by two dermatologists. Length of intervention: not stated. Length of follow-up: not stated. How intervention was assessed: participants completed a baseline survey, tests of their triage abilities when using and not using the software, and an exit survey | Participants made fewer incorrect decisions when using the software (36.7% versus 13.3%) | Only a pilot study; small sample; selection bias; no evidence of change in clinical practice | 4 stars |
| Gerbert et al, 200221 | US | Primary care physicians | RCT | Skin | To determine whether a skin cancer triage intervention could be effectively delivered over the internet | Intervention components: tutorial modules: registration, pre-test, pre-test scores with individualised feedback, skin cancer instruction, post-test, post-test II, exit survey. participants had unlimited time to complete the modules but could not redo a module once completed. Number of participants recruited: 879. Number of participants completing study: 27 in the intervention group and 19 in the control group completed the whole programme. Intervention delivered by: online. Length of intervention: N/A, online. Length of follow-up: 8 weeks between post-tests I and II. How intervention was assessed: the pre-test module included 36 images of skin lesions (cancerous and non-cancerous), for which participants had to select a diagnosis and evaluation plan. Individualised feedback was then given. Post-test I contained a set of 25 different lesions. Post-test II contained the same lesions as given in post-test I | Intervention participants significantly improved on overall diagnosis and evaluation planning, and in diagnosing malignant melanoma; only malignant melanoma accuracy persisted at 8-week follow-up (correctly diagnosed 88% pre-test and 90% at post-test II) | High dropout rate; no set amount of time for participants to complete modules — may have been able to look up information, therefore giving false results; Problem with randomisation — differences between control and intervention participants; small clinical effect | |
| Girgis et al, 199522 | Australia | Family doctors | RCT | Skin | To evaluate the effectiveness of a postgraduate training programme in improving family doctors' levels of knowledge and their clinical practice in relation to skin cancer diagnosis and management | Intervention components: split into three sessions; epidemiology, diagnosis and management (lecture session), melanoma unit (accompanying a specialist surgeon in follow-up examinations), and practical surgical procedures. Number of participants recruited: 65. Number of participants completing study: 41. Intervention delivered by: members of the local melanoma unit. Length of intervention: not stated—first two sessions were 3 hours each. Length of follow-up: not stated. How intervention was assessed: mailed survey assessing confidence and knowledge; a set of coloured slides and case histories to test diagnosis and management; patient self-report and pathology data were checked to assess increases in screening rates; and proportion of attempted diagnoses made | Significant improvements were made in the proportion of accurate diagnoses, correct management of test lesions, and confidence of doctors to provide advice to patients | Lack of detail about the length of time between intervention and follow-up; poor randomisation process; selection bias; no evidence of implications for practice; small sample size and evidence that the study is not representative of all GPs | 6 stars |
| Jiwa et al, 200623 | UK | General practices (not clear if GPs only or other healthcare professionals were included) | RCT | Colorectal cancer | To examine whether the introduction of an electronic interactive referral pro forma or educational outreach visits by a colorectal surgeon to general practice can alter the case mix of patients referred to lower-bowel specialists | Intervention components: practices were randomised to receive either an educational outreach visit, an interactive electronic pro forma for processing referrals, both, or neither. The outreach visit involved a surgeon delivering short educational sessions tailored to the needs of the target audience, which summarised the features of significant colorectal disease and encouraged questions. The electronic forms requested information from drop-down menus for 15 signs and symptoms previously identified by GPs and surgeons as predicting colorectal disease. Once clinical data were entered, a referral letter was automatically produced, seeking an appropriate appointment at a hospital clinic. Number of participants recruited: 44 practices (180GPs). Number of participants completing study: not stated (assumed all recruited). Intervention delivered by: colorectal surgeons. Length of intervention: 6 months. Length of follow-up: 6 months. How intervention was assessed: an ‘assessment score’ was given for the quality of referral letters and the proportion of patients referred. Interviews were carried outwith participants who were assigned the electronic referral, but no information is given for the other groups | No evidence that either intervention was successful and did not increase the proportion of patients with pathology who were referred; pro forma documented better assessment of patients | Little information on assessment of participants; only 18% of GPs actually used the software; no a assessment of knowledge before and after the educational visit to assess impact | 7 stars |
| Khan, 200924 | UK | GPs | Feasibility study | Colorectal cancer | To test the feasibility of a paper -based assessment tool incorporating the CAPER score (a clinical prediction rule for patients presenting to primary care with lower-gastrointestinal symtpoms) | Intervention components: a paper-based assessment tool was developed using the CAPER score and NICE guidelines for colorectal cancer. The GP had to record the patients' symptomhistory, clinical and rectal examinations, full blood count (FBC), and faecal occult blood (FOB) (if no overt bleeding), and then follow the referral advice. Number of participants recruited: 122. Number of participants completing study: 122. Intervention delivered by: GPs. Length of intervention: maximum of 3 months. Time to followup: 3 months. How intervention was assessed: three audits were conducted during and after the intervention to obtain consultation data and to check if the assessment tool was being correctly filled in. The final audit collected the clinical outcomes and final diagnoses of patients | Recruitment rateswere poor, with only 24% of the recruitment target being achieved. Rates were higher for GP than reception staff recruitment. As directed by the assesment tool, 93% of patients had a clinical examination but only 64% a rectal examination. 48% had a FBC and 38% FOB tests. In only 55% were the CAPER scores correctly calculated | Inadequate amount of follow-up; poor recruitment; selection bias; not powered to detect clinical effect | 4 stars |
| Kirklin et al, 200725 | England | GPs and primary care nurses | RCT | Skin | To investigate whether the observational skills of doctors and nurses can beimproved by arts-based observational skills training | Intervention components: all practices received an on-site educational outreach programme. The intervention group received practical observational skills training delivered by an artist, while the control group had a lecture on the aetiology and management of psoriasis and acne and applying emollients, delivered by a dermatologist. Number of participants recruited: 68. Number of participants completing study: 68. Intervention delivered by: artist (intervention), dermatologist (control). Length of intervention: 90 minutes, plus 30 minutes for testing. Length of follow-up: not stated. How intervention was assessed: the test consisted of participants describing a series of three dermatological photographs (diagnosis was not an outcome) | The intervention appeared to have a significant effect on improving observational skills in the post-intervention test; the authors acknowledge that the study did not express how useful such skills would be in practice | Not clear if any follow-up was carried out; participants were not randomised; selection bias | 6 stars |
| Logan et al, 200226 | England | General practices | RCT | Bowel | To assess the adequacy of investigation of iron deficiency anaemia and to establish whether a simple computer-generated prompt would increase the completeness of treatment and investigation of patients presenting in general practice | Intervention components: practices were randomised after being stratified by district and practice size. Patients were identified fromtheir blood indices—laboratory computers were programmed to print an appropriate prompt (control prompt: ‘consistent with iron deficiency: ?cause’, and intervention prompt: ‘consistent with iron deficiency:- ?cause, suggest treat with ferrous sulphate, 200mg t.i.d for 4 months but check response in 3–4 weeks. Simultaneously investigate cause. Consider barium enema to exclude colorectal problems’). Number of participants recruited: 603. Number of participants completing study: 431 patients included in analysis. Intervention delivered by:Hospital pathologists. Length of intervention: 12 months. Length of follow-up: 12 months. How intervention was assessed: patient records were analysed for adequate management of anaemia instigated within 3 months of symptom presentation, as defined by four criteria. | 47% of patients were managed adequately, but the prompt was not found to affect the level of investigation or adequacy of follow-up. There was an increase in iron therapy | No ‘no prompt’ control group; cross contamination (GPs in control and intervention groups may influence each other); outcome measure based on secondary care management guidelines | 9 stars |
| Moreno-Ramirez et al, 200527 | Spain | GPs | Before-and-after study | Skin | To compare diagnosis and management options after the evaluation of clinical and dermatoscopic teleconsultations with a store-and-forward teledermatology screening system for pigmented lesions | Intervention components: all patients attending a participating primary care centre who met the eligibility criteria were enrolled into the study. Digital clinical pictures and dermatoscopic pictures were taken of the patients and submitted to the hospital. All patients underwent normal face-to-face consultations and skin biopsy. Number of participants recruited: 63 patients. Number of participants completing study: 61 patients. Intervention delivered by: GPs. Length of intervention: 1 month. Length of follow-up: 4 months. How intervention was assessed: two participant dermatologists from the PLC evaluated teleconsultations without dermatoscopic images, and completed a form listing the possible diagnoses, their confidence level, management decision, and whether the patient should be referred at two time points during the study. They then were re-assessed 2 months later using the same information with dermoscopic images | Dermoscopic teleconsultation yielded more relevant referrals than the clinical teleconsultation, and GPs spent more time on the dermoscopic teleconsultation. Without dermascopy referral was recommended in 47.5% of cases as opposed to 39.3% with dermascopy; no false negatives | Small sample size; short follow-up; selection not randomor consecutive; not felt to be cost-effective | 4 stars |
| Nekhyludov et al, 200811 | US | Primary care clinicians (physicians, nurse practitioners, certified nurse midwives, and physician assistants) | Before-and-after study | Breast | Pilot test and evaluation of an office-based intervention aimed at improving outcomes by increasing the use of breast-symptom-related guidelines | Intervention components: sites were split into intervention and controls (no information about how this was done). At the intervention sites, at every visit of a female patient with breast symptoms, clinicians were sent a packet of materials containing two guidelines and a patient information sheet. Control sites were informed of the materials but were not given any directly. Number of participants recruited: 123. Number of participants completing intervention: 101. Intervention delivered by: not stated. Length of intervention: 8 months. Length of follow-up: not stated. How intervention was assessed: surveys to measure knowledge of breast symptoms based on information provided by the guidelines were administered before and after the intervention | The intervention was found to increase guideline use, but had no statistically significant effect on knowledge and attitudes | Short follow-up; not randomised—participants were mainly female which makes generalising the results more difficult; small sample size; no objective evidence that the guidelines were actually used | 4 stars |
| Raasch et al, 200028 | Australia | Family physicians | RCT | Skin | Would practitioners' management of skin cancer improve if feedback highlighted specific deficiencies in their diagnosis and performance of skin excisions and grouped peer data for comparison? | Intervention components: doctors obtained patients' consent for an audit of excised skin lesions. Information from this was then fed back to the doctors and included an educational message (intervention group). The control group received no feedback. Number of participants recruited: 46. Number of participants completing intervention: 41. Intervention delivered by: not stated. Length of assessment: 9 months. Length of follow-up: 3 months. How intervention were assessed: audits were conducted before and after the feedback | Doctors in the intervention group improved their recordings of the clinical diagnosis on the pathology request form. There was no significant improvement in correct diagnosis; significant differences were found between the intervention and control group at baseline | High dropout rate; small sample size; possible contamination between intervention and control groups; poor randomisation | 9 stars |
| Shah et al, 200629 | Egypt and Tunisia | Primary care physicians | Before-and-after study | Breast | To improve knowledge of epidemiology and management of breast cancer cases | Intervention components: participants were invited to a conference, which consisted of a pre-test session, a presentation of an educational module (30-minute presentation and question and answer session by a local oncologist) and post-testing. Number of participants recruited: 194. Number of participants completing intervention: not stated. Intervention delivered by: local oncologist. Length of assessment: not stated. Length of follow-up: none (immediate post-test). How intervention was assessed: the test questions covered topics included in the educational module, focusing on epidemiology, management, and referral in a true/false format | Significant improvement in knowledge after the intervention, but for many questions there was no significant change and Tunisian participants generally performed worse then Egyptian participants | Selection bias; not clear who presented the educational module (the oncologist or a member of the study team); no follow-up to see how long the effects lasted | 1 star |
| Soliman et al, 200630 | Pakistan | Primary care physicians | Before-and-after study | Breast | To measure the effect of a continuing medical education breast cancer module on knowledge of primary care physicians | Intervention components: PowerPoint-based educational module that included breast cancer risk factors, early detection by clinical breast examand appropriate referral. Number of participants recruited: 183. Number of participants completing intervention: 133. Intervention delivered by: not stated. Length of intervention: not stated. Length of follow-up: none (immediate post-test). How intervention was assessed: tests were before and after the presentation, and included questions on current knowledge of several types of breast cancer, in a true/false format | Participants showed improvement for all knowledge questions following the intervention. The study controlled for postgraduate education and number of breast cancer cases seen in the previous year | No follow-up period to assess knowledge retention; small sample; selection bias | 1 star |
| Westerhoff et al, 200031 | Australia | Primary care physicians | RCT | Skin (malignant melanoma) | To determine whether primary care physicians could improve their melanoma diagnosis using surface microscopy after a short educational intervention | Intervention components: the intervention group was given an atlas that described a method for diagnosing invasive melanomas, and a 1-hour presentation on surface microscopy that reviewed the diagnostic method in the atlas and included a quiz. Number of participants recruited: 74. Number of participants completing intervention: not stated. Programme delivered by: not stated. Length of intervention: not stated. Length of follow-up: median time 23 days. How intervention was assessed: the tests included 100 multiple-choice photos of skin lesions | Significant improvement in diagnostic ability post-intervention (pre-intervention 54.6% correct, post-education 62.7% correct, post-education and surface microscopy 75.9%) | Small sample size; selection bias; short follow-up; range of 2–54 days between completing intervention and completing post-test, no set time, may have affected results | 5 stars |
| Wolters et al, 200432 | The Netherlands | GPs | RCT | Prostate | To examine the effects of a distance-learning programme on patient outcomes | Intervention components: the intervention group received amultifaceted distance learning programme: (1) a package for individual learning (PIL), (2) consultation support materials including a voiding diary, the International Prostate Symptom Score and Bother Score, (3) the Dutch lower urinary tract symptoms (LUTS) guideline summarised into a decision tree (4) 2 PILs (one on PSA testing and one on LUTS treatments). The PIL consisted of a knowledge test—answers were sent to a central institute and a set of standard correct answers was returned as feedback. Number of participants: 142. Number of participants completing intervention: 89. Intervention delivered by: N/A. Length of intervention: not stated. Length of follow-up: not stated. How intervention was assessed: via the questionnaire. | PSA testing requests were higher in the intervention group (not a desired study outcome); patients' fear of cancer was more of a motivating factor for GPs. Intervention did decrease the number of urology referrals. This was not expected, as the decision-support tree was felt to suggest a low rate of testing. | High dropout rate; small sample size | 8 stars |
| You et al, 200733 | Australia | GPs | Before-and-after study | Skin | To evaluate a self-instructional education module with audit and feedback, designed to increase the skills of GPs in diagnosing melanocytic lesions and skin cancer | Intervention components: education package to assist in diagnosis and management of skin cancers and improve knowledge of the management and treatment of suspicious lesions, and procedures and guidelines for management and follow-up. Number of participants recruited: 16. Number of participants completing intervention: not stated (assumed all). Intervention delivered by: not stated. Length of intervention: 18 months. Length of follow-up: 6 months. How intervention was assessed: a baseline clinical self-audit of skin excisions was carried out over 6 months, followed by a feedback report and provision of the educational module over the next 6 months. A post-education self-audit was then redone for the last 6 months | Intervention appeared to improve the malignant: benign ratio for excised malanocytic lesions—double the number of malignant melanomas were diagnosed | Small sample size, not randomly selected; high dropout rate 3 stars | |
N/A = not available. QA = quality assessment. PLC = pigmented lesion clinic.