INTRODUCTION

Guidelines recommend primary care physicians (PCPs) offer patients a choice of colorectal cancer (CRC) screening methods, including colonoscopy and fecal occult blood tests (FOBT).1 However, in countries like the USA and Switzerland, patients are screened almost exclusively with colonoscopy.2, 3 When offered both tests, patients appear as likely to choose one as the other; the predominance of colonoscopy may largely be explained by physician preference and local medical culture.4 Offering only colonoscopy might explain why screening rates are low.

We sought to determine the proportions of patients who opted for screening with colonoscopy or FOBT and who refused testing among 50–75 year olds eligible for screening at a PCP visit. We described variation in care between PCPs and tried to identify PCP-level factors associated with testing method and refusal.

METHODS

We conducted a cross-sectional data collection on CRC screening practices at PCP level. We invited 129 PCPs from the Swiss Sentinel Surveillance Network (Sentinella) to fill a structured data collection form for 40 consecutive non-emergent consultations with 50–75-year-old patients. The federal office of public health (FOPH) provided demographic data at PCP level. PCPs reported demographic data at patient-level, data on previous CRC tests, contraindications for screening, risk factors for CRC, if CRC screening was discussed, choice of test (colonoscopy, FOBT, other), and refusal for testing. We calculated overall proportions and reported variation between PCPs in the proportion of FOBT vs. colonoscopy they prescribed to patients who chose to be tested. We calculated overall prescription rates of FOBT vs. colonoscopy for each PCP, including both patients who had already undergone screening and patients prescribed screening after the consultation. We dichotomized this covariate by never-prescription of FOBT (no patients previously tested with FOBT or prescribed FOBT after discussion) vs. any FOBT.

We used mixed-effects logistic regression models that allowed us to cluster the data by PCP (with PCPs modeled as a random effect) to explore the association between PCP characteristics and the proportion of patients who refused screening after discussion. We adjusted the models for PCPs’ demographics (age, sex) and language region, for patients’ demographics (age, sex), and PCPs’ prescription patterns.

RESULTS

Ninety-one PCPs (71% of invited, mean age 54, 24% women) collected data on 3637 patients. One hundred eighty-six patients were excluded because they were not aged 50–75 years old or had already been seen during data collection. The analysis included 3453 patients (mean age 63, 50% women). PCPs discussed screening with 51% (874/1727) of eligible patients (not up-to-date and no contraindications for testing) (Fig. 1). After excluding patients with risk factors or symptoms suggestive of CRC (n = 104), 61% (473/770) opted for screening (FOBT/colonoscopy ratio 0.5), 29% refused, 6% were undecided, and 3% were unspecified or missing. Most patients who refused screening said they did so because they did not feel concerned.

Figure 1
figure 1

Flowchart of 40 consecutive patients aged 50–75 included by PCPs from the Sentinella network in 2017. PCPs collected data on 40 consecutive patients aged 50–75 from on past screening status, contraindications for screening, if a discussion on CRC screening could take place, RF and symptoms for CRC, and the decision taken (refusal, FOBT, colonoscopy, other). Data collected between April and December 2017. a RF, risk factor for CRC.

Full size image

Thirty-three PCPs (36%) had none of their patients previously tested with FOBT or who planned to be tested with FOBT. Patients of PCPs who only offered colonoscopy were more likely to refuse screening than patients of PCPs who offered both colonoscopy and FOBT (44% vs. 20%, respectively, Fig. 2). These results were confirmed in our mixed-effects multivariate model (OR 3.90, 95%CI 1.90 to 8.00, p < 0.001). No other PCP characteristics were associated with chosen testing methods or refusal rates.

Figure 2
figure 2

Decision patterns among patients who had a discussion on CRC screening ( N patients = 770) and included by PCPs who only prescribed colonoscopy ( N = 33) vs. PCPs who prescribed both colonoscopy and FOBT ( N = 58), in the Sentinella Network in 2017. Patients with risk factors or symptoms suggestive for CRC ( n = 104) (see Fig. 1 ) excluded of this analysis.

Full size image

DISCUSSION

When PCPs discussed CRC screening with their 50–75-year-old patients who were not up-to-date with screening, had no contraindication, and no risk factors for CRC, a third of their patients declined to be screened. PCPs who only offered colonoscopy had lower screening rates (47% vs. 71%) and higher refusal rates (44% vs. 20%) than PCPs who offered both colonoscopy and FOBT. These results are in line with a randomized controlled trial showing lower uptake rates of CRC screening tests among patients who are offered only colonoscopy vs. among the ones who are offered both FOBT and colonoscopy.5 We were inherently limited in considering additional patient-level sociodemographic factors by the simplicity and anonymity of our data collection.

Encouraging PCPs to offer both methods could reduce the number of physicians who only prescribe one screening modality, reduce variation between practices, and allow more patients to choose the test that matches their preferences and values.4, 6 This could reduce the number of refusals, raise CRC screening rates, and ultimately lower the burden of CRC.