Survey
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
Association between use of the English urgent referral pathway for suspected cancer and mortality outcome in cancer patients: cohort study Henrik Møller, Carolynn Gildea, David Meechan, Greg Rubin, Thomas Round, Peter Vedsted Cancer Epidemiology and Population Health, KCL (HM) Public Health England, Knowledge & Intelligence (CG, DM, HM) Durham University (GR) Primary Care and Public Health Sciences, KCL (TR) Cancer Diagnosis in Primary Care, Aarhus University (HM, PV) Cancer in the UK • >1 in 2 people in the UK with develop cancer (Ahmad 2015) • >300,000 patients are diagnosed and >150,000 die every year from cancer in the UK (CRUK) • Late diagnosis is a major contributing factor to relatively poorer cancer survival rates in the UK (Richards 2009) • >10,000 premature cancer deaths per year could be avoided if survival rates matched the best in Europe (AbdelRahman 2007) • Nearly 1 in 4 patients diagnosed via an emergency admission route with poorer 1 year survival (Elliss Brookes 2012) ICBP: 1 year relative survival 90 Coleman et al, Lancet 2011 45 AUS CAN SWE NOR DEN SWE UK AUS 85 CAN 40 SWE AUS CAN 80 NOR NOR 35 DEN DEN 75 UK UK 30 25 70 20 65 1995-99 2000-02 1995-99 2005-07 Colorectal Cancer 1yr RS 2000-02 2005-07 Lung Cancer 1yr RS 100 80 SWE 98 NOR 75 CAN AUS CAN AUS 96 70 DEN NOR DEN 94 UK 65 UK 60 92 55 90 1995-99 Breast CancerCAN 1yr RS AUS 2000-02 SWE 2005-07 NOR DEN UK 50 1995-99 AUS CANRS Ovarian Cancer 1yr 2000-02 NOR DEN 2005-07 UK Delays in Symptomatic Cancer Diagnosis Model of Total Patient Delay (Walter 2012) Cancer: National Awareness and Early Diagnosis Initiative (NAEDI) (Hiom 2015) Primary Care and Cancer • Most of those with cancer present with symptoms, and most of these presentations are to primary care (Rubin 2011) • Suspecting a diagnosis of cancer in general practice is not straightforward: average full time GP 7-8 new cancer cases/year (Richards 2009) • Low predictive value of even “red flag” symptoms eg PPV rectal bleeding 2.4% (Jones 2007) • Many patients have multiple GP visits prior to referral especially those with cancers of greater diagnostic difficulty (Lyratzopoulos 2014) Primary Care and Cancer Three or more general practitioner consultations before hospital referral (Lyratzopoulos 2012) Referral for suspected cancer • Referral guidelines for suspected cancer, DoH 2000 • Referral guidelines for suspected cancer, NICE 2005 ”Urgent referral” (Two week wait/2ww) • The patient is seen by a specialist within 14 days, and • The patient is treated within 62 days from referral or within 31 days from the decision to treat • Updated NICE referral guidance due for publication May 2015 Change in diagnostic intervals 2001/02 – 2007/08 (Neal 2013) The “waiting time paradox” (Torring 2011) Urgent Referrals 2009-2014 Use of the urgent referral pathway and mortality outcome in cancer patients Variation in a care process ? Variation in a relevant outcome 10,000 0 5,000 Frequency 15,000 Histogram of referral ratio 0 1 2 Referral ratio ratio Referral 3 4 20,000 0 10,000 Frequency 30,000 40,000 Histogram of conversion rate 0 .2 .4 .6 PPV rate Conversion .8 1 10,000 0 5,000 Frequency 15,000 Histogram of detection rate 0 .2 .4 .6 Sensitivity Detection rate .8 1 Strategy of analysis Mortality of urgently referred cancer patients ? = Mortality of other cancer patients Strategy of analysis Mortality of urgently referred cancer patients Mortality of cancer patients from GP practices with high use of urgent referral ? = Mortality of other cancer patients ? = Mortality of cancer patients from GP practices with low use of urgent referral Data: England; 2009; 215,284 cancer patients from 8049 GP practices r: 0.42 Analysis • Cohort analysis of death rates, adjusted for age, sex and cancer type • Cox proportional hazards regression • Sensitivity analyses Analysis by tertiles of distributions Table 1. Hazard ratios for death in relation to referral ratio, conversion rate and detection rate. Variable Referral ratio Tertile Median Persons Deaths 0.68 1.01 1.39 71773 71768 71743 31136 30417 30067 1.05 1.00 0.97 <0.001 1.04 1.07 0.96 0.99 1 2 3 0.08 0.12 0.17 71811 72101 71372 30206 30672 30742 1.00 1.00 1.01 0.117 0.99 1.02 1.00 1.03 1 2 3 0.33 0.43 0.54 71804 72065 71415 31072 30749 29799 1.04 1.00 0.96 <0.001 1.02 1.06 0.95 0.98 trend p-value Detection rate 95% CI 1 2 3 trend p-value Conversion rate HR trend p-value *HR adjusted for age, sex and cancer type Table 1. Hazard ratios for death in relation to referral ratio, conversion rate and detection rate. Variable Referral ratio Tertile Median Persons Deaths 0.68 1.01 1.39 71773 71768 71743 31136 30417 30067 1.05 1.00 0.97 <0.001 1.04 1.07 0.96 0.99 1 2 3 0.08 0.12 0.17 71811 72101 71372 30206 30672 30742 1.00 1.00 1.01 0.117 0.99 1.02 1.00 1.03 1 2 3 0.33 0.43 0.54 71804 72065 71415 31072 30749 29799 1.04 1.00 0.96 <0.001 1.02 1.06 0.95 0.98 trend p-value Detection rate 95% CI 1 2 3 trend p-value Conversion rate HR trend p-value *HR adjusted for age, sex and cancer type Analysis by deciles of distributions ratioby fordecile death,of byreferral decile ofratio. referral ratio distribution A. Hazard ratio Hazard for death, 1.05 Hazard ratio 1.00 0.95 0.90 0.85 1 2 3 4 5 6 Decile 7 8 9 10 for death, by decile ofrate. PPV distribution B. Hazard ratio for Hazard death, ratio by decile of conversion 1.10 Hazard ratio 1.05 1.00 0.95 0.90 1 2 3 4 5 6 Decile 7 8 9 10 Hazard ratio death, decile of rate. sensitivity distribution C. Hazard ratio for death, byfor decile ofby detection 1.05 Hazard ratio 1.00 0.95 0.90 0.85 1 2 3 4 5 6 Decile 7 8 9 10 Joint effects of referral ratio, conversion rate and detection rate Table 2. Hazard ratios for death in 3-way classification of referral ratio, conversion rate and detection rate tertiles. Numbers of persons and deaths, and hazard ratios for death with 95% confidence intervals. The lower part of the table shows the three aggregate groups suggested by the analysis. Referral ratio low Referral ratio intermediate Detection rate low Conversion rate low Conversion rate intermediate Conversion rate high 11347 4906 1.08 11512 5011 1.07 15123 6622 1.06 Detection rate intermediate 1.12 971 421 1.08 1.03 1.11 4343 1874 1.03 1.03 15538 6898 1.10 1.06 1.04 Hazard ratios adjusted for age, sex and cancer type. 0.98 0.98 1.03 Detection rate high 1.19 340 126 0.87 1.08 1711 699 0.98 1.10 10888 4579 1.00 0.73 0.91 0.96 Detection rate low 1.03 11627 4961 1.03 1.06 7903 3423 1.02 1.04 1052 528 1.22 Referral ratio high Detection rate intermediate 6252 2575 0.98 1.07 0.97 13622 5738 1.06 1.00 8679 3620 0.97 1.12 7594 3152 0.97 13269 5673 1.01 0.98 0.93 1.03 1770 747 1.05 0.99 1.34 0.94 Detection rate high 0.97 0.93 0.95 Detection rate low 1.13 11824 4982 1.00 1.02 1253 562 1.07 1.02 163 77 1.18 Detection rate intermediate 0.96 14338 6002 1.04 0.98 0.98 1.17 8137 3505 0.99 1.47 1270 584 1.10 0.94 Detection rate high 0.94 13342 5486 1.02 0.96 0.92 0.99 0.95 14941 6240 1.03 0.96 0.93 1.00 1.01 6475 2629 0.92 0.88 0.96 1.19 Referral ratio high Detection rate low 11824 4982 1.00 1253 562 1.07 163 77 1.18 Detection rate intermediate 0.96 14338 6002 1.04 0.98 0.98 1.17 8137 3505 0.99 1.47 1270 584 1.10 0.94 Detection rate high 0.94 13342 5486 1.02 0.96 0.92 0.99 0.95 14941 6240 1.03 0.96 0.93 1.00 1.01 6475 2629 0.92 0.88 0.96 1.19 Referral ratio high 11824 4982 1.00 1253 562 1.07 163 77 1.18 0.96 14338 6002 0.98 1.04 0.98 1.17 8137 3505 0.99 1.47 1270 584 1.10 0.94 Detection rate high Detection rate intermediate Detection rate low 0.94 13342 5486 0.96 1.02 0.92 0.99 0.95 14941 6240 0.96 1.03 0.93 1.00 1.01 6475 2629 0.92 0.88 0.96 1.19 Table 2. Hazard ratios for death in 3-way classification of referral ratio, conversion rate and detection rate tertiles. Numbers of persons and deaths, and hazard ratios for death with 95% confidence intervals. The lower part of the table shows the three aggregate groups suggested by the analysis. Referral ratio low Referral ratio intermediate Detection rate low Conversion rate low Conversion rate intermediate Conversion rate high 11347 4906 1.08 11512 5011 1.07 15123 6622 1.06 Detection rate intermediate 1.12 971 421 1.08 1.03 1.11 4343 1874 1.03 1.03 15538 6898 1.10 1.06 1.04 Hazard ratios adjusted for age, sex and cancer type. 0.98 0.98 1.03 Detection rate high 1.19 340 126 0.87 1.08 1711 699 0.98 1.10 10888 4579 1.00 0.73 0.91 0.96 Detection rate low 1.03 11627 4961 1.03 1.06 7903 3423 1.02 1.04 1052 528 1.22 Referral ratio high Detection rate intermediate 6252 2575 0.98 1.07 0.97 13622 5738 1.06 1.00 8679 3620 0.97 1.12 7594 3152 0.97 13269 5673 1.01 0.98 0.93 1.03 1770 747 1.05 0.99 1.34 0.94 Detection rate high 0.97 0.93 0.95 Detection rate low 1.13 11824 4982 1.00 1.02 1253 562 1.07 1.02 163 77 1.18 Detection rate intermediate 0.96 14338 6002 1.04 0.98 0.98 1.17 8137 3505 0.99 1.47 1270 584 1.10 0.94 Detection rate high 0.94 13342 5486 1.02 0.96 0.92 0.99 0.95 14941 6240 1.03 0.96 0.93 1.00 1.01 6475 2629 0.92 0.88 0.96 1.19 Referral ratio low Detection rate low Conversion rate low Conversion rate intermediate Conversion rate high 11347 4906 1.08 11512 5011 1.07 15123 6622 1.06 Detection rate intermediate 1.12 971 421 1.08 1.03 1.11 4343 1874 1.03 1.03 15538 6898 1.10 1.06 1.04 0.98 0.98 1.03 Detection rate high 1.19 340 126 0.87 0.73 1.03 1.08 1711 699 0.98 0.91 1.06 1.10 10888 4579 1.00 0.96 1.04 Referral ratio low Detection rate low Conversion rate low Conversion rate intermediate Conversion rate high 11347 4906 1.08 11512 5011 1.07 15123 6622 1.06 Detection rate intermediate 1.12 971 421 1.08 1.03 1.11 4343 1874 1.03 1.03 15538 6898 1.10 1.06 1.04 0.98 0.98 1.03 Detection rate high 1.19 340 126 0.87 0.73 1.03 1.08 1711 699 0.98 0.91 1.06 1.10 10888 4579 1.00 0.96 1.04 Table 2. Hazard ratios for death in 3-way classification of referral ratio, conversion rate and detection rate tertiles. Numbers of persons and deaths, and hazard ratios for death with 95% confidence intervals. The lower part of the table shows the three aggregate groups suggested by the analysis. Referral ratio low Referral ratio intermediate Detection rate low Conversion rate low Conversion rate intermediate Conversion rate high 11347 4906 1.08 11512 5011 1.07 15123 6622 1.06 Detection rate intermediate 1.12 971 421 1.08 1.03 1.11 4343 1874 1.03 1.03 15538 6898 1.10 1.06 1.04 Hazard ratios adjusted for age, sex and cancer type. 0.98 0.98 1.03 Detection rate high 1.19 340 126 0.87 1.08 1711 699 0.98 1.10 10888 4579 1.00 0.73 0.91 0.96 Detection rate low 1.03 11627 4961 1.03 1.06 7903 3423 1.02 1.04 1052 528 1.22 Referral ratio high Detection rate intermediate 6252 2575 0.98 1.07 0.97 13622 5738 1.06 1.00 8679 3620 0.97 1.12 7594 3152 0.97 13269 5673 1.01 0.98 0.93 1.03 1770 747 1.05 0.99 1.34 0.94 Detection rate high 0.97 0.93 0.95 Detection rate low 1.13 11824 4982 1.00 1.02 1253 562 1.07 1.02 163 77 1.18 Detection rate intermediate 0.96 14338 6002 1.04 0.98 0.98 1.17 8137 3505 0.99 1.47 1270 584 1.10 0.94 Detection rate high 0.94 13342 5486 1.02 0.96 0.92 0.99 0.95 14941 6240 1.03 0.96 0.93 1.00 1.01 6475 2629 0.92 0.88 0.96 1.19 A. Overall analysis (215,284 patients; 91,620 deaths) Group* 1 High 2 3 Low p-trend Patients 34,758 121,692 58,834 % 16 57 27 Deaths 14,355 51,533 25,732 HR 0.95 1.00 1.07 <0.001 95% CI 0.94 0.97 1.05 1.08 *Group 1: High referral ratio and high detection rate. Group 2: All other combinations than group 1 and group 3. Group 3: Low referral ratio and low or intermediate detection rate. Hazard ratios adjusted for age, sex and cancer type. B. Restricted to colorectal cancer Group 1 High 2 3 Low p-trend Patients 3,865 13,982 6,789 % 16 57 28 Deaths 1,670 6,247 3,073 HR 0.96 1.00 1.03 0.032 95% CI 0.91 1.02 0.98 1.07 Deaths 3,107 11,173 5,656 HR 0.96 1.00 1.04 <0.001 95% CI 0.92 1.00 1.01 1.08 Deaths 997 3,366 1,678 HR 1.07 1.00 1.09 0.247 95% CI 1.00 1.15 1.03 1.16 Deaths 984 3,362 1,703 HR 0.95 1.00 1.08 0.001 95% CI 0.88 1.02 1.02 1.15 C. Restricted to lung cancer Group 1 High 2 3 Low p-trend Patients 3,596 12,792 6,398 % 16 56 28 D. Restricted to female breast cancer Group 1 High 2 3 Low p-trend Patients 5,108 19,236 9,225 % 15 57 27 E. Restricted to prostate cancer Group 1 High 2 3 Low p-trend Patients 4,325 14,259 7,071 % 17 56 28 L. Not adjusted for type of cancer Group Patients 1 High 34,758 2 121,692 3 Low 58,834 p-trend % 16 57 27 Deaths 14,355 51,533 25,732 HR 0.96 1.00 1.07 <0.001 95% CI 0.94 0.97 1.05 1.08 J. Also adjusted for the patient's socioeconomic status** Group Patients 1 34,758 2 121,692 3 58,834 p-trend % 16 57 27 Deaths 14,355 51,533 25,732 HR 0.95 1.00 1.06 <0.001 **From J: Most vs. least deprived quintile: HR: 1.35 (1.32-1.37); p-trend over quintiles <0.001 95% CI 0.94 0.97 1.05 1.08 Underlying mechanisms? High use of urgent referral Low mortality; (High survival) Underlying mechanisms? High use of urgent referral Short delay; Earlier stage; Treatment options Low mortality; (High survival) Underlying mechanisms? High use of urgent referral GP’s awareness of cancer ? Short delay; Earlier stage; Treatment options Low mortality; (High survival) Underlying mechanisms? High use of urgent referral GP’s awareness of cancer ? Short delay; Earlier stage; Treatment options Low mortality; (High survival) M. Also adjusted for GP population age and sex distribution*** Group Patients 1 High 34,758 2 121,692 3 Low 58,834 p-trend % 16 57 27 Deaths 14,355 51,533 25,732 HR 0.96 1.00 1.06 <0.001 95% CI 0.94 0.98 1.04 1.07 ***From M: Effects of GP population characteristics: HR: 0.88 (0.86-0.90) for highest vs. lowest tertile of median age Conclusions and implications • The analysis suggests that urgent referral for suspected cancer is efficacious. • The roles of a direct mechanism (via reduced delay) and indirect mechanisms (via awareness) are not known. • GP practices with consistent low use of urgent referral may benefit their cancer patients from using it more often, and from being more aware of signs of cancer. r: 0.42 Detection rate r: 0.42 Referral ratio Table 2. Hazard ratios for death in 3-way classification of referral ratio, conversion rate and detection rate tertiles. Numbers of persons and deaths, and hazard ratios for death with 95% confidence intervals. The lower part of the table shows the three aggregate groups suggested by the analysis. Referral ratio low Referral ratio intermediate Detection rate low Conversion rate low Conversion rate intermediate Conversion rate high 11347 4906 1.08 11512 5011 1.07 15123 6622 1.06 Detection rate intermediate 1.12 971 421 1.08 1.03 1.11 4343 1874 1.03 1.03 15538 6898 1.10 1.06 1.04 Group 3; 27% HR: 1.07 (1.05-1.08) 0.98 0.98 1.03 Detection rate high 1.19 340 126 0.87 1.08 1711 699 0.98 1.10 10888 4579 1.00 0.73 0.91 0.96 Detection rate low 1.03 11627 4961 1.03 1.06 7903 3423 1.02 1.04 1052 528 1.22 Group 2; 57% HR: 1.00 Referral ratio high Detection rate intermediate 6252 2575 0.98 1.07 0.97 13622 5738 1.06 1.00 8679 3620 0.97 1.12 7594 3152 0.97 13269 5673 1.01 0.98 0.93 1.03 1770 747 1.05 0.99 1.34 0.94 Detection rate high 0.97 0.93 0.95 Detection rate low 1.13 11824 4982 1.00 1.02 1253 562 1.07 1.02 163 77 1.18 Detection rate intermediate 0.96 14338 6002 1.04 0.98 0.98 1.17 8137 3505 0.99 1.47 1270 584 1.10 0.94 Detection rate high 0.94 13342 5486 1.02 0.96 0.92 0.99 0.95 14941 6240 1.03 0.96 0.93 1.00 1.01 6475 2629 0.92 0.88 0.96 1.19 Group 1; 16% HR: 0.95 (0.94-0.97) Hazard ratios adjusted for age, sex and cancer type. 27% of patients: 1.07 (1.05-1.08) 57% of patients 1.00 16% of patients 0.95 (0.94-0.97) Table 2. Hazard ratios for death in 3-way classification of referral ratio, conversion rate and detection rate tertiles. Numbers of persons and deaths, and hazard ratios for death with 95% confidence intervals. The lower part of the table shows the three aggregate groups suggested by the analysis. Referral ratio low Referral ratio intermediate Detection rate low Conversion rate low Conversion rate intermediate Conversion rate high 11347 4906 1.08 11512 5011 1.07 15123 6622 1.06 Detection rate intermediate 1.12 971 421 1.08 1.03 1.11 4343 1874 1.03 1.03 15538 6898 1.10 1.06 1.04 Hazard ratios adjusted for age, sex and cancer type. 0.98 0.98 1.03 Detection rate high 1.19 340 126 0.87 1.08 1711 699 0.98 1.10 10888 4579 1.00 0.73 0.91 0.96 Detection rate low 1.03 11627 4961 1.03 1.06 7903 3423 1.02 1.04 1052 528 1.22 Referral ratio high Detection rate intermediate 6252 2575 0.98 1.07 0.97 13622 5738 1.06 1.00 8679 3620 0.97 1.12 7594 3152 0.97 13269 5673 1.01 0.98 0.93 1.03 1770 747 1.05 0.99 1.34 0.94 Detection rate high 0.97 0.93 0.95 Detection rate low 1.13 11824 4982 1.00 1.02 1253 562 1.07 1.02 163 77 1.18 Detection rate intermediate 0.96 14338 6002 1.04 0.98 0.98 1.17 8137 3505 0.99 1.47 1270 584 1.10 0.94 Detection rate high 0.94 13342 5486 1.02 0.96 0.92 0.99 0.95 14941 6240 1.03 0.96 0.93 1.00 1.01 6475 2629 0.92 0.88 0.96 1.19 Table 2. Hazard ratios for death in 3-way classification of referral ratio, conversion rate and detection rate tertiles. Numbers of persons and deaths, and hazard ratios for death with 95% confidence intervals. The lower part of the table shows the three aggregate groups suggested by the analysis. Referral ratio low Referral ratio intermediate Detection rate low Conversion rate low Conversion rate intermediate Conversion rate high 11347 4906 1.08 11512 5011 1.07 15123 6622 1.06 Detection rate intermediate 1.12 971 421 1.08 1.03 1.11 4343 1874 1.03 1.03 15538 6898 1.10 1.06 1.04 Hazard ratios adjusted for age, sex and cancer type. 0.98 0.98 1.03 Detection rate high 1.19 340 126 0.87 1.08 1711 699 0.98 1.10 10888 4579 1.00 0.73 0.91 0.96 Detection rate low 1.03 11627 4961 1.03 1.06 7903 3423 1.02 1.04 1052 528 1.22 Referral ratio high Detection rate intermediate 6252 2575 0.98 1.07 0.97 13622 5738 1.06 1.00 8679 3620 0.97 1.12 7594 3152 0.97 13269 5673 1.01 0.98 0.93 1.03 1770 747 1.05 0.99 1.34 0.94 Detection rate high 0.97 0.93 0.95 Detection rate low 1.13 11824 4982 1.00 1.02 1253 562 1.07 1.02 163 77 1.18 Detection rate intermediate 0.96 14338 6002 1.04 0.98 0.98 1.17 8137 3505 0.99 1.47 1270 584 1.10 0.94 Detection rate high 0.94 13342 5486 1.02 0.96 0.92 0.99 0.95 14941 6240 1.03 0.96 0.93 1.00 1.01 6475 2629 0.92 0.88 0.96 1.19 Table 2. Hazard ratios for death in 3-way classification of referral ratio, conversion rate and detection rate tertiles. Numbers of persons and deaths, and hazard ratios for death with 95% confidence intervals. The lower part of the table shows the three aggregate groups suggested by the analysis. Referral ratio low Referral ratio intermediate Detection rate low Conversion rate low Conversion rate intermediate Conversion rate high 11347 4906 1.08 11512 5011 1.07 15123 6622 1.06 Detection rate intermediate 1.12 971 421 1.08 1.03 1.11 4343 1874 1.03 1.03 15538 6898 1.10 1.06 1.04 Group 3; 27% HR: 1.07 (1.05-1.08) Hazard ratios adjusted for age, sex and cancer type. 0.98 0.98 1.03 Detection rate high 1.19 340 126 0.87 1.08 1711 699 0.98 1.10 10888 4579 1.00 0.73 0.91 0.96 Group 2; 57% HR: 1.00 Detection rate low 1.03 11627 4961 1.03 1.06 7903 3423 1.02 1.04 1052 528 1.22 Referral ratio high Detection rate intermediate 6252 2575 0.98 1.07 0.97 13622 5738 1.06 1.00 8679 3620 0.97 1.12 7594 3152 0.97 13269 5673 1.01 0.98 0.93 1.03 1770 747 1.05 0.99 1.34 0.94 Detection rate high 0.97 0.93 0.95 Detection rate low 1.13 11824 4982 1.00 1.02 1253 562 1.07 1.02 163 77 1.18 Detection rate intermediate 0.96 14338 6002 1.04 0.98 0.98 1.17 8137 3505 0.99 1.47 1270 584 1.10 0.94 Detection rate high 0.94 13342 5486 1.02 0.96 0.92 0.99 0.95 14941 6240 1.03 0.96 0.93 1.00 1.01 6475 2629 0.92 0.88 0.96 1.19 Group 1; 16% HR: 0.95 (0.94-0.97)