Cancer incidence in patients with a high normal platelet count: a cohort study using primary care data

Cancer incidence in patients with a high normal platelet count: a cohort study using primary care... Abstract Background A platelet count >400 × 109/l (i.e. thrombocytosis) is a recently discovered risk marker of cancer. The risk of undiagnosed cancer in patients with thrombocytosis is 11.6% for men and 6.2% for women, well above the 3% risk threshold set by National Institute for Health and Care Excellence (NICE) for cancer investigation. Patients with a platelet count at the upper end of the normal range (325–400 × 109/l) could be at increased risk of undiagnosed malignancy. Objective To quantify the risk of an undiagnosed cancer in patients with a platelet count at the upper end of the normal range. Methods A primary care-based cohort study using Clinical Practice Research Datalink (CPRD) data from 2000 to 2013. The study sample comprised 2704 individuals stratified by platelet count: 325–349 × 109/l; 350–374 × 109/l; 375–399 × 109/l. Incident cancer diagnoses in the year following that platelet count were obtained from patient records. Results Cancer incidence rose with increasing platelet count: 2.6% [95% confidence interval (CI) 1.9 to 3.6] in subjects with a count of 325–349 × 109/l, 3.7% (95% CI 2.5 to 5.3) in subjects with a count of 350–374 × 109/l and 5.1% (95% CI 3.4 to 7.5) in those with a count of 375–399 × 109/l. Colorectal cancer was most commonly diagnosed in all three groups. Cancer incidence was consistently higher in males than in females. Conclusion These results suggest that clinicians should consider cancer in patients with a platelet count >375 × 109/l, review reasons for testing and any additional reported symptoms. Until these results are replicated on a larger scale, recommendations for clinical action cannot be made. Cancer epidemiology, diagnosis, incidence, platelet count, primary care, risk marker Introduction Cancer survival in the UK is improving, but lags behind that in other European countries (1,2). Improving earlier diagnosis has been identified as a key strategy to improve survival (1). A range of research and policy initiatives has aimed to achieve this including public awareness campaigns and 2-week wait clinics. A valuable approach to achieving earlier diagnosis is to identify signs and symptoms that are associated with underlying malignancy to help GPs select patients for referral for definitive diagnostic testing. Previous approaches to improving cancer diagnosis have included the production of risk assessment tools (RATs) which present the risk of cancer associated with various clinical signs and symptoms. One such sign is thrombocytosis (3) (high platelet count, usually >400 × 109/l although lower values have been proposed (4)). The positive predictive value of thrombocytosis for detecting any cancer is 11.6% [95% confidence interval (CI) 11.0 to 12.3] in men, and 6.2% (95% CI 5.9 to 6.5) in women (3). This risk value far exceeds the 3% threshold set by the UK National Institute for Health and Care Excellence (NICE) at which patients are recommended for referral for possible cancer (5). In that study, even mildly elevated platelet counts had a positive predictive value for cancer above 3%. Patients with a platelet count at the upper end of the normal range may also be at increased risk of cancer; identifying these patients in primary care may be the first ‘clue’ to an undiagnosed malignancy which could help to achieve earlier diagnosis, before other symptoms have developed. In this study, we aimed to quantify the risk of undiagnosed cancer in patients with a platelet count at the upper end of the normal range. Methods Data sources Electronic medical records from the Clinical Practice Research Datalink (CPRD, www.cprd.co.uk) linked with English National Cancer Registration Service (NCRAS) data (6). The CPRD compiles patient records from UK primary care, holding data on approximately 7% of the UK population. The NCRAS gathers patient data from screening and imaging services, secondary care patient administration systems and hospital episode statistics. Study population The study sample was a randomly selected 10000 from the CPRD database (the comparison group from our earlier study (3)) who met the following inclusion criteria: First platelet count from 2000 to 2013 was from 150 to 399 × 109/l Aged ≥40 years at the time of the platelet count No cancer diagnosis recorded prior to that platelet count Exclusion criteria included: Aged under 40 years at index date Diagnosed with non-melanoma skin cancer after index date (commonly under recorded) From this sample, patients who had a platelet count from 325 to 399 × 109/l were selected for the present study. The first qualifying platelet count in the study period was designated the ‘index date’. Subjects were stratified into three subgroups: Group 1: 325–349 × 109/l Group 2: 350–374 × 109/l Group 3: 375–399 × 109/l These ranges were chosen to be narrow enough to enable clinically useful indicators of when a platelet count should prompt further action for suspected cancer, while being wide enough to maintain reasonable sample sizes. Subjects were excluded if they had a cancer record prior to their platelet count index date. Study outcomes and analyses New cancer diagnoses were determined by searching CPRD records in the year after the platelet index date for any of 2134 cancer-related codes, organized into 23 sites and from NCRAS records; the earliest record was taken as the date of diagnosis. The one-year cancer incidence (and 95% CI) was estimated for Groups 1–3. The cancer site was identified; where more than one site was recorded, the earliest record was taken as the primary site, and only one cancer diagnosis was recorded per individual. A chi-square test of independence was performed to compare the frequency of cancer diagnosis in each of the three groups. Cancer incidence data were stratified by sex in each group. It is not possible to determine from CPRD data why blood tests were ordered. Patients’ records in the 3 weeks before the index date were searched for codes for single symptoms that should prompt urgent referral in NICE guidance (NG12) (5). The proportion of subjects with these ‘alarms’ symptoms in the 21 days before their blood test was determined, and compared for subjects with and without a subsequent cancer diagnosis. Stata version 14.2 was used to execute all analyses (7). This paper conforms to STROBE reporting guidelines (8). Results The study sample included 2704 individuals after exclusions (Fig. 1). Group 1 (platelet count 325–349 × 109/l) included 1439 subjects [328 (22.8%) males]. The median age at the index date was 69.4 years [interquartile range (IQR): 58.3–79.2]. Cancer was diagnosed in 38 patients within one year, an incidence of 2.6% (95% CI 1.9 to 3.6) (Table 1). Colorectal (n = 7; 18%) and lung (n = 5; 13%) were the most commonly recorded cancers (Fig. 2). Table 1. Number of cancers diagnosed in each platelet count group during follow-up and the cancer incidence Group  Platelet count range (×109/l)  Subjects  Number diagnosed with cancer within 1 year  One year incidence % (95% CI)  1  325–349  1439  38  2.6 (1.9 to 3.6)  2  350–374  779  29  3.7 (2.5 to 5.3)  3  375–399  486  25  5.1 (3.4 to 7.5)  Group  Platelet count range (×109/l)  Subjects  Number diagnosed with cancer within 1 year  One year incidence % (95% CI)  1  325–349  1439  38  2.6 (1.9 to 3.6)  2  350–374  779  29  3.7 (2.5 to 5.3)  3  375–399  486  25  5.1 (3.4 to 7.5)  Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. View Large Table 1. Number of cancers diagnosed in each platelet count group during follow-up and the cancer incidence Group  Platelet count range (×109/l)  Subjects  Number diagnosed with cancer within 1 year  One year incidence % (95% CI)  1  325–349  1439  38  2.6 (1.9 to 3.6)  2  350–374  779  29  3.7 (2.5 to 5.3)  3  375–399  486  25  5.1 (3.4 to 7.5)  Group  Platelet count range (×109/l)  Subjects  Number diagnosed with cancer within 1 year  One year incidence % (95% CI)  1  325–349  1439  38  2.6 (1.9 to 3.6)  2  350–374  779  29  3.7 (2.5 to 5.3)  3  375–399  486  25  5.1 (3.4 to 7.5)  Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. View Large Figure 1. View largeDownload slide Patient flow diagram to show the number of subjects included in each of the platelet count groups and the number excluded from the original study sample. Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. Figure 1. View largeDownload slide Patient flow diagram to show the number of subjects included in each of the platelet count groups and the number excluded from the original study sample. Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. Figure 2. View largeDownload slide Sites of cancer diagnoses for (a) male and (b) female patients diagnosed with cancer in each of the three platelet count groups. Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. Figure 2. View largeDownload slide Sites of cancer diagnoses for (a) male and (b) female patients diagnosed with cancer in each of the three platelet count groups. Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. Group 2 (350–374 × 109/l) included 779 subjects [164 (21.1%) males]. The median age at index date was 72.0 (IQR: 59.2–80.9). Cancer was diagnosed in 29 patients within one year, an incidence of 3.7% (95% CI 2.5 to 5.3) (Table 1). Colorectal (n = 9; 31%) and lung (n = 3; 10%) cancers were also the most commonly diagnosed in this cohort (Fig. 2). Group 3 (375–399 × 109/l) included 486 subjects [118 (24.3%) males]. The median age at index date was 71.7 (IQR: 58.9–81.3). Cancer was diagnosed in 25 patients within one year, an incidence of 5.1% (95% CI 3.4 to 7.5) (Table 1). Colorectal was the most commonly diagnosed cancer in this cohort (n = 7; 28%), followed by lung and prostate cancers (for both, n = 2; 7%) (Fig. 2). When the groups were stratified by sex, the cancer incidence was consistently higher in men than in women (Table 2). Table 2. Numbers of cancers diagnosed in each platelet count group during follow-up and the cancer incidence for that group, by sex Group  Sex  Subjects  Number diagnosed with cancer within one year  One year incidence, % (95% CI)  1  Men  328  15  4.6 (2.6 to 7.4)  Women  1111  23  2.1 (1.3 to 3.1)  2  Men  164  12  7.3 (3.8 to 12.4)  Women  615  17  2.8 (1.6 to 4.4)  3  Men  118  10  8.5 (4.1 to 15.0)  Women  368  15  4.1 (2.3 to 6.6)  Group  Sex  Subjects  Number diagnosed with cancer within one year  One year incidence, % (95% CI)  1  Men  328  15  4.6 (2.6 to 7.4)  Women  1111  23  2.1 (1.3 to 3.1)  2  Men  164  12  7.3 (3.8 to 12.4)  Women  615  17  2.8 (1.6 to 4.4)  3  Men  118  10  8.5 (4.1 to 15.0)  Women  368  15  4.1 (2.3 to 6.6)  Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. View Large Table 2. Numbers of cancers diagnosed in each platelet count group during follow-up and the cancer incidence for that group, by sex Group  Sex  Subjects  Number diagnosed with cancer within one year  One year incidence, % (95% CI)  1  Men  328  15  4.6 (2.6 to 7.4)  Women  1111  23  2.1 (1.3 to 3.1)  2  Men  164  12  7.3 (3.8 to 12.4)  Women  615  17  2.8 (1.6 to 4.4)  3  Men  118  10  8.5 (4.1 to 15.0)  Women  368  15  4.1 (2.3 to 6.6)  Group  Sex  Subjects  Number diagnosed with cancer within one year  One year incidence, % (95% CI)  1  Men  328  15  4.6 (2.6 to 7.4)  Women  1111  23  2.1 (1.3 to 3.1)  2  Men  164  12  7.3 (3.8 to 12.4)  Women  615  17  2.8 (1.6 to 4.4)  3  Men  118  10  8.5 (4.1 to 15.0)  Women  368  15  4.1 (2.3 to 6.6)  Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. View Large The chi-square test found a significant relationship between platelet count and cancer diagnosis: χ2(2; N = 2714) = 26.3, P ≤ 0.01. Patients with higher platelet counts were more likely to be diagnosed with cancer than those with lower platelet counts. In the 21 days before the index test, alarm symptoms (5) were recorded for 47 of the 2704 (1.7%) patients. The proportion reporting an alarm symptom was greater in patients who developed cancer within 1 year of the index test compared with those who did not; 9/92 (9.8%; 95% CI 4.6 to 17.8) versus 38/2612 (1.5%; 95% CI 1.0 to 2.0) respectively. Discussion Cancer incidence increased with increasing platelet count, exceeding 3% in subjects with a platelet count of 375–399 × 109/l, and being consistently higher in men than in women. Baseline platelet levels may be higher or benign causes of raised platelets more likely, in women than in men. The proportion of males in the sample was just above 20%; this supports the former suggestion. The influence of sex on platelet count is poorly understood, and worthy of further research. Colorectal was the most commonly diagnosed cancer type; this is in contrast with our previous work on cancer incidence in patients with a platelet count of >400 × 109/l, in which lung was most commonly diagnosed (3). In that study, a much higher proportion of lung and colorectal cancers were diagnosed than would be expected given national incidence data (and a much lower proportion of breast and prostate cancers). In the present study, too few cancers were diagnosed in the sample to make similar comparisons. Comparison with existing literature This is the first study to consider cancer risk with a platelet count in the high normal range. All previous studies have used a threshold of ≥400 × 109/l when examining the clinical utility of platelet count in diagnosing cancer. In a recent systematic review, thrombocytosis was found to be an independent predictor of four types of cancer in studies of single cancer sites: lung, kidney, oesophago-gastric and uterine cancer (9). Cancer incidence rose with increasing platelet count. Despite the small sample size, it appears that the relationship between cancer risk and platelet counts is monotonic, and begins well in the ‘normal’ range. The concept of a single threshold defining normality is semi-arbitrary, and based on distributions in the healthy population. Our study population—primary care patients who had a full blood count taken—has presumably more ill health than the general population. There are many and varied clinical reasons for taking a full blood count; indeed, at least a quarter of the adult population in any one year has this test (10). Although our data are from a selected population, having a blood test is unlikely to introduce any bias specifically towards patients who are suspected of having an undiagnosed cancer due to the wide range of reasons for testing. Therefore, it is very unlikely this effect can explain our results, particularly with the clear dose–response effect. Males had consistently higher cancer risk than females. In a recent study, revised upper limits for platelet count were proposed to be 362 × 109/l for males and 405 × 109/l for females, supporting our finding that a platelet count in the 350–400 × 109/l range should be more worthy of concern in males than in females (4). If our findings are replicated, it is likely that the threshold platelet count warranting consideration of cancer will be lower in males than in females. Strengths and limitations This study uses a robust data source, the CPRD, which has been used extensively in past studies of cancer risk markers (11–14). The use of NCRAS data is a further strength, identifying incident cases that may have been unrecorded in the CPRD. Blood counts are electronically transmitted to the CPRD, reducing the risk of recording error. This study is based on a convenience sample of patients taken from a previous study (3), resulting in small sample sizes and wide confidence intervals for the risk estimates. The reasons why the blood tests were ordered in the sample are unknown; cancer may have been suspected prior to blood testing. Cancer alarm symptoms accounted for a negligible proportion (1.7%) of all symptoms in the 3 weeks before the index test, suggesting that cancer was not suspected in the vast majority of patients having blood testing. Group 1 had a lower median age than the other two cohorts (69.4 in Group 1, 72.0 and 71.7 in Groups 2 and 3, respectively) which may have had an impact on the lower proportion of cancer diagnoses in that group. The sample in the present study was too small to investigate the effect of age on the relationship between platelet count and cancer diagnosis; future work should address this. Conclusions This study is small, but suggests that the risk of cancer in men with platelet counts >325 × 109/l exceeds 3%. For women, the figure is 375 × 109/l, though for platelet values in the range 350–374 × 109/l the risk is 2.8% (95% CI 1.6 to 4.4), still above the level at which patients would like investigation (15). This suggests that clinicians should consider a cancer diagnosis in patients with a platelet count above these values. This could lead to earlier diagnosis, potentially at an earlier disease stage, if the patient is referred for further investigation sooner than they would have been had the raised platelet count not been recognized as a risk marker. A clinician receiving a high-normal platelet count should review why the test was done and what ongoing symptoms the patient is reporting. This finding is currently only a clue towards possible cancer. Until the findings are replicated in a much larger sample, and the specific cancers delineated in greater detail with data on stage at diagnosis, a blanket recommendation for investigation would be premature. Declaration Funding: The Policy Research Unit in Cancer Awareness, Screening and Early Diagnosis receives funding for a research programme from the Department of Health Policy Research Programme. It is a collaboration between researchers from seven institutions (Queen Mary University of London, University College London [UCL], King’s College London, London School of Hygiene and Tropical Medicine, Hull York Medical School, Durham University and the University of Exeter). OCU is supported by the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care South West Peninsula. The views expressed are those of the author(s) and not necessarily those of the National Health Service (NHS), the National Institute for Health Research (NIHR) or the Department of Health. Ethical approval: Ethical committee approval was given by the Independent Scientific Advisory Committee of the Clinical Practice Research Datalink: reference number 13-007. Conflict of interests: WH is an associate editor of Family Practice. The authors declare no other competing interests. References 1. Coleman MP, Forman D, Bryant H, et al.   Cancer survival in Australia, Canada, Denmark, Norway, Sweden, and the UK, 1995–2007 (the international cancer benchmarking partnership): an analysis of population-based cancer registry data. Lancet  2011; 377: 127– 38. Google Scholar CrossRef Search ADS PubMed  2. Walters S, Benitez-Majano S, Muller P, et al.   Is England closing the international gap in cancer survival? Br J Cancer  2015; 113: 848– 60. Google Scholar CrossRef Search ADS PubMed  3. Bailey SE, Ukoumunne OC, Shephard EA, Hamilton W. Clinical relevance of thrombocytosis in primary care: a prospective cohort study of cancer incidence using English electronic medical records and cancer registry data. Br J Gen Pract  2017; 67: e405– 13. Google Scholar CrossRef Search ADS PubMed  4. Biino G, Santimone I, Minelli C, et al.   Age- and sex-related variations in platelet count in Italy: a proposal of reference ranges based on 40987 subjects’ data. PLoS One  2013; 8: e54289. Google Scholar CrossRef Search ADS PubMed  5. NICE. Suspected cancer: recognition and referral [NG12]. 2015. http://www.nice.org.uk/guidance/NG12 (accessed on 5 September 2017). 6. Herrett E, Gallagher AM, Bhaskaran K, et al.   Data resource profile: Clinical Practice Research Datalink (CPRD). Int J Epidemiol  2015; 44: 827– 36. Google Scholar CrossRef Search ADS PubMed  7. StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC. 8. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol  2008; 61: 344– 9. Google Scholar CrossRef Search ADS PubMed  9. Bailey SE, Ukoumunne OC, Shephard E, Hamilton W. How useful is thrombocytosis in predicting an underlying cancer in primary care? A systematic review. Fam Pract  2017; 34: 4– 10. Google Scholar CrossRef Search ADS PubMed  10. Hamilton W, Lancashire R, Sharp D, Peters TJ, Cheng KK, Marshall T. The importance of anaemia in diagnosing colorectal cancer: a case—control study using electronic primary care records. Br J Cancer  2008; 98: 323– 7. Google Scholar CrossRef Search ADS PubMed  11. Shephard EA, Neal RD, Rose PW, Walter FM, Hamilton W. Symptoms of adult chronic and acute leukaemia before diagnosis: large primary care case-control studies using electronic records. Br J Gen Pract  2016; 66: e182– 8. Google Scholar CrossRef Search ADS PubMed  12. Shephard EA, Neal RD, Rose P, Walter FM, Litt EJ, Hamilton WT. Quantifying the risk of multiple myeloma from symptoms reported in primary care patients: a large case-control study using electronic records. Br J Gen Pract  2015; 65: e106– 13. Google Scholar CrossRef Search ADS PubMed  13. McGlynn EA, McDonald KM, Cassel CK. Measurement is essential for improving diagnosis and reducing diagnostic error: a report from the Institute of Medicine. JAMA  2015; 314: 2501– 2502. http://www. nationalacademies.org/hmd/Activities/Quality/Diagnostic ErrorHealthCare.aspx. 14. Hamilton F, Carroll R, Hamilton W, Salisbury C. The risk of cancer in primary care patients with hypercalcaemia: a cohort study using electronic records. Br J Cancer  2014; 111: 1410– 2. Google Scholar CrossRef Search ADS PubMed  15. Banks J, Hollinghurst S, Bigwood L, Peters TJ, Walter FM, Hamilton W. Preferences for cancer investigation: a vignette-based study of primary-care attendees. Lancet Oncol  2014; 15: 232– 40. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Family Practice Oxford University Press

Cancer incidence in patients with a high normal platelet count: a cohort study using primary care data

Loading next page...
 
/lp/ou_press/cancer-incidence-in-patients-with-a-high-normal-platelet-count-a-VYzXhzC99i
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0263-2136
eISSN
1460-2229
D.O.I.
10.1093/fampra/cmy018
Publisher site
See Article on Publisher Site

Abstract

Abstract Background A platelet count >400 × 109/l (i.e. thrombocytosis) is a recently discovered risk marker of cancer. The risk of undiagnosed cancer in patients with thrombocytosis is 11.6% for men and 6.2% for women, well above the 3% risk threshold set by National Institute for Health and Care Excellence (NICE) for cancer investigation. Patients with a platelet count at the upper end of the normal range (325–400 × 109/l) could be at increased risk of undiagnosed malignancy. Objective To quantify the risk of an undiagnosed cancer in patients with a platelet count at the upper end of the normal range. Methods A primary care-based cohort study using Clinical Practice Research Datalink (CPRD) data from 2000 to 2013. The study sample comprised 2704 individuals stratified by platelet count: 325–349 × 109/l; 350–374 × 109/l; 375–399 × 109/l. Incident cancer diagnoses in the year following that platelet count were obtained from patient records. Results Cancer incidence rose with increasing platelet count: 2.6% [95% confidence interval (CI) 1.9 to 3.6] in subjects with a count of 325–349 × 109/l, 3.7% (95% CI 2.5 to 5.3) in subjects with a count of 350–374 × 109/l and 5.1% (95% CI 3.4 to 7.5) in those with a count of 375–399 × 109/l. Colorectal cancer was most commonly diagnosed in all three groups. Cancer incidence was consistently higher in males than in females. Conclusion These results suggest that clinicians should consider cancer in patients with a platelet count >375 × 109/l, review reasons for testing and any additional reported symptoms. Until these results are replicated on a larger scale, recommendations for clinical action cannot be made. Cancer epidemiology, diagnosis, incidence, platelet count, primary care, risk marker Introduction Cancer survival in the UK is improving, but lags behind that in other European countries (1,2). Improving earlier diagnosis has been identified as a key strategy to improve survival (1). A range of research and policy initiatives has aimed to achieve this including public awareness campaigns and 2-week wait clinics. A valuable approach to achieving earlier diagnosis is to identify signs and symptoms that are associated with underlying malignancy to help GPs select patients for referral for definitive diagnostic testing. Previous approaches to improving cancer diagnosis have included the production of risk assessment tools (RATs) which present the risk of cancer associated with various clinical signs and symptoms. One such sign is thrombocytosis (3) (high platelet count, usually >400 × 109/l although lower values have been proposed (4)). The positive predictive value of thrombocytosis for detecting any cancer is 11.6% [95% confidence interval (CI) 11.0 to 12.3] in men, and 6.2% (95% CI 5.9 to 6.5) in women (3). This risk value far exceeds the 3% threshold set by the UK National Institute for Health and Care Excellence (NICE) at which patients are recommended for referral for possible cancer (5). In that study, even mildly elevated platelet counts had a positive predictive value for cancer above 3%. Patients with a platelet count at the upper end of the normal range may also be at increased risk of cancer; identifying these patients in primary care may be the first ‘clue’ to an undiagnosed malignancy which could help to achieve earlier diagnosis, before other symptoms have developed. In this study, we aimed to quantify the risk of undiagnosed cancer in patients with a platelet count at the upper end of the normal range. Methods Data sources Electronic medical records from the Clinical Practice Research Datalink (CPRD, www.cprd.co.uk) linked with English National Cancer Registration Service (NCRAS) data (6). The CPRD compiles patient records from UK primary care, holding data on approximately 7% of the UK population. The NCRAS gathers patient data from screening and imaging services, secondary care patient administration systems and hospital episode statistics. Study population The study sample was a randomly selected 10000 from the CPRD database (the comparison group from our earlier study (3)) who met the following inclusion criteria: First platelet count from 2000 to 2013 was from 150 to 399 × 109/l Aged ≥40 years at the time of the platelet count No cancer diagnosis recorded prior to that platelet count Exclusion criteria included: Aged under 40 years at index date Diagnosed with non-melanoma skin cancer after index date (commonly under recorded) From this sample, patients who had a platelet count from 325 to 399 × 109/l were selected for the present study. The first qualifying platelet count in the study period was designated the ‘index date’. Subjects were stratified into three subgroups: Group 1: 325–349 × 109/l Group 2: 350–374 × 109/l Group 3: 375–399 × 109/l These ranges were chosen to be narrow enough to enable clinically useful indicators of when a platelet count should prompt further action for suspected cancer, while being wide enough to maintain reasonable sample sizes. Subjects were excluded if they had a cancer record prior to their platelet count index date. Study outcomes and analyses New cancer diagnoses were determined by searching CPRD records in the year after the platelet index date for any of 2134 cancer-related codes, organized into 23 sites and from NCRAS records; the earliest record was taken as the date of diagnosis. The one-year cancer incidence (and 95% CI) was estimated for Groups 1–3. The cancer site was identified; where more than one site was recorded, the earliest record was taken as the primary site, and only one cancer diagnosis was recorded per individual. A chi-square test of independence was performed to compare the frequency of cancer diagnosis in each of the three groups. Cancer incidence data were stratified by sex in each group. It is not possible to determine from CPRD data why blood tests were ordered. Patients’ records in the 3 weeks before the index date were searched for codes for single symptoms that should prompt urgent referral in NICE guidance (NG12) (5). The proportion of subjects with these ‘alarms’ symptoms in the 21 days before their blood test was determined, and compared for subjects with and without a subsequent cancer diagnosis. Stata version 14.2 was used to execute all analyses (7). This paper conforms to STROBE reporting guidelines (8). Results The study sample included 2704 individuals after exclusions (Fig. 1). Group 1 (platelet count 325–349 × 109/l) included 1439 subjects [328 (22.8%) males]. The median age at the index date was 69.4 years [interquartile range (IQR): 58.3–79.2]. Cancer was diagnosed in 38 patients within one year, an incidence of 2.6% (95% CI 1.9 to 3.6) (Table 1). Colorectal (n = 7; 18%) and lung (n = 5; 13%) were the most commonly recorded cancers (Fig. 2). Table 1. Number of cancers diagnosed in each platelet count group during follow-up and the cancer incidence Group  Platelet count range (×109/l)  Subjects  Number diagnosed with cancer within 1 year  One year incidence % (95% CI)  1  325–349  1439  38  2.6 (1.9 to 3.6)  2  350–374  779  29  3.7 (2.5 to 5.3)  3  375–399  486  25  5.1 (3.4 to 7.5)  Group  Platelet count range (×109/l)  Subjects  Number diagnosed with cancer within 1 year  One year incidence % (95% CI)  1  325–349  1439  38  2.6 (1.9 to 3.6)  2  350–374  779  29  3.7 (2.5 to 5.3)  3  375–399  486  25  5.1 (3.4 to 7.5)  Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. View Large Table 1. Number of cancers diagnosed in each platelet count group during follow-up and the cancer incidence Group  Platelet count range (×109/l)  Subjects  Number diagnosed with cancer within 1 year  One year incidence % (95% CI)  1  325–349  1439  38  2.6 (1.9 to 3.6)  2  350–374  779  29  3.7 (2.5 to 5.3)  3  375–399  486  25  5.1 (3.4 to 7.5)  Group  Platelet count range (×109/l)  Subjects  Number diagnosed with cancer within 1 year  One year incidence % (95% CI)  1  325–349  1439  38  2.6 (1.9 to 3.6)  2  350–374  779  29  3.7 (2.5 to 5.3)  3  375–399  486  25  5.1 (3.4 to 7.5)  Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. View Large Figure 1. View largeDownload slide Patient flow diagram to show the number of subjects included in each of the platelet count groups and the number excluded from the original study sample. Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. Figure 1. View largeDownload slide Patient flow diagram to show the number of subjects included in each of the platelet count groups and the number excluded from the original study sample. Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. Figure 2. View largeDownload slide Sites of cancer diagnoses for (a) male and (b) female patients diagnosed with cancer in each of the three platelet count groups. Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. Figure 2. View largeDownload slide Sites of cancer diagnoses for (a) male and (b) female patients diagnosed with cancer in each of the three platelet count groups. Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. Group 2 (350–374 × 109/l) included 779 subjects [164 (21.1%) males]. The median age at index date was 72.0 (IQR: 59.2–80.9). Cancer was diagnosed in 29 patients within one year, an incidence of 3.7% (95% CI 2.5 to 5.3) (Table 1). Colorectal (n = 9; 31%) and lung (n = 3; 10%) cancers were also the most commonly diagnosed in this cohort (Fig. 2). Group 3 (375–399 × 109/l) included 486 subjects [118 (24.3%) males]. The median age at index date was 71.7 (IQR: 58.9–81.3). Cancer was diagnosed in 25 patients within one year, an incidence of 5.1% (95% CI 3.4 to 7.5) (Table 1). Colorectal was the most commonly diagnosed cancer in this cohort (n = 7; 28%), followed by lung and prostate cancers (for both, n = 2; 7%) (Fig. 2). When the groups were stratified by sex, the cancer incidence was consistently higher in men than in women (Table 2). Table 2. Numbers of cancers diagnosed in each platelet count group during follow-up and the cancer incidence for that group, by sex Group  Sex  Subjects  Number diagnosed with cancer within one year  One year incidence, % (95% CI)  1  Men  328  15  4.6 (2.6 to 7.4)  Women  1111  23  2.1 (1.3 to 3.1)  2  Men  164  12  7.3 (3.8 to 12.4)  Women  615  17  2.8 (1.6 to 4.4)  3  Men  118  10  8.5 (4.1 to 15.0)  Women  368  15  4.1 (2.3 to 6.6)  Group  Sex  Subjects  Number diagnosed with cancer within one year  One year incidence, % (95% CI)  1  Men  328  15  4.6 (2.6 to 7.4)  Women  1111  23  2.1 (1.3 to 3.1)  2  Men  164  12  7.3 (3.8 to 12.4)  Women  615  17  2.8 (1.6 to 4.4)  3  Men  118  10  8.5 (4.1 to 15.0)  Women  368  15  4.1 (2.3 to 6.6)  Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. View Large Table 2. Numbers of cancers diagnosed in each platelet count group during follow-up and the cancer incidence for that group, by sex Group  Sex  Subjects  Number diagnosed with cancer within one year  One year incidence, % (95% CI)  1  Men  328  15  4.6 (2.6 to 7.4)  Women  1111  23  2.1 (1.3 to 3.1)  2  Men  164  12  7.3 (3.8 to 12.4)  Women  615  17  2.8 (1.6 to 4.4)  3  Men  118  10  8.5 (4.1 to 15.0)  Women  368  15  4.1 (2.3 to 6.6)  Group  Sex  Subjects  Number diagnosed with cancer within one year  One year incidence, % (95% CI)  1  Men  328  15  4.6 (2.6 to 7.4)  Women  1111  23  2.1 (1.3 to 3.1)  2  Men  164  12  7.3 (3.8 to 12.4)  Women  615  17  2.8 (1.6 to 4.4)  3  Men  118  10  8.5 (4.1 to 15.0)  Women  368  15  4.1 (2.3 to 6.6)  Group 1: 325–349 × 109/l. Group 2: 350–374 × 109/l. Group 3: 375–399 × 109/l. View Large The chi-square test found a significant relationship between platelet count and cancer diagnosis: χ2(2; N = 2714) = 26.3, P ≤ 0.01. Patients with higher platelet counts were more likely to be diagnosed with cancer than those with lower platelet counts. In the 21 days before the index test, alarm symptoms (5) were recorded for 47 of the 2704 (1.7%) patients. The proportion reporting an alarm symptom was greater in patients who developed cancer within 1 year of the index test compared with those who did not; 9/92 (9.8%; 95% CI 4.6 to 17.8) versus 38/2612 (1.5%; 95% CI 1.0 to 2.0) respectively. Discussion Cancer incidence increased with increasing platelet count, exceeding 3% in subjects with a platelet count of 375–399 × 109/l, and being consistently higher in men than in women. Baseline platelet levels may be higher or benign causes of raised platelets more likely, in women than in men. The proportion of males in the sample was just above 20%; this supports the former suggestion. The influence of sex on platelet count is poorly understood, and worthy of further research. Colorectal was the most commonly diagnosed cancer type; this is in contrast with our previous work on cancer incidence in patients with a platelet count of >400 × 109/l, in which lung was most commonly diagnosed (3). In that study, a much higher proportion of lung and colorectal cancers were diagnosed than would be expected given national incidence data (and a much lower proportion of breast and prostate cancers). In the present study, too few cancers were diagnosed in the sample to make similar comparisons. Comparison with existing literature This is the first study to consider cancer risk with a platelet count in the high normal range. All previous studies have used a threshold of ≥400 × 109/l when examining the clinical utility of platelet count in diagnosing cancer. In a recent systematic review, thrombocytosis was found to be an independent predictor of four types of cancer in studies of single cancer sites: lung, kidney, oesophago-gastric and uterine cancer (9). Cancer incidence rose with increasing platelet count. Despite the small sample size, it appears that the relationship between cancer risk and platelet counts is monotonic, and begins well in the ‘normal’ range. The concept of a single threshold defining normality is semi-arbitrary, and based on distributions in the healthy population. Our study population—primary care patients who had a full blood count taken—has presumably more ill health than the general population. There are many and varied clinical reasons for taking a full blood count; indeed, at least a quarter of the adult population in any one year has this test (10). Although our data are from a selected population, having a blood test is unlikely to introduce any bias specifically towards patients who are suspected of having an undiagnosed cancer due to the wide range of reasons for testing. Therefore, it is very unlikely this effect can explain our results, particularly with the clear dose–response effect. Males had consistently higher cancer risk than females. In a recent study, revised upper limits for platelet count were proposed to be 362 × 109/l for males and 405 × 109/l for females, supporting our finding that a platelet count in the 350–400 × 109/l range should be more worthy of concern in males than in females (4). If our findings are replicated, it is likely that the threshold platelet count warranting consideration of cancer will be lower in males than in females. Strengths and limitations This study uses a robust data source, the CPRD, which has been used extensively in past studies of cancer risk markers (11–14). The use of NCRAS data is a further strength, identifying incident cases that may have been unrecorded in the CPRD. Blood counts are electronically transmitted to the CPRD, reducing the risk of recording error. This study is based on a convenience sample of patients taken from a previous study (3), resulting in small sample sizes and wide confidence intervals for the risk estimates. The reasons why the blood tests were ordered in the sample are unknown; cancer may have been suspected prior to blood testing. Cancer alarm symptoms accounted for a negligible proportion (1.7%) of all symptoms in the 3 weeks before the index test, suggesting that cancer was not suspected in the vast majority of patients having blood testing. Group 1 had a lower median age than the other two cohorts (69.4 in Group 1, 72.0 and 71.7 in Groups 2 and 3, respectively) which may have had an impact on the lower proportion of cancer diagnoses in that group. The sample in the present study was too small to investigate the effect of age on the relationship between platelet count and cancer diagnosis; future work should address this. Conclusions This study is small, but suggests that the risk of cancer in men with platelet counts >325 × 109/l exceeds 3%. For women, the figure is 375 × 109/l, though for platelet values in the range 350–374 × 109/l the risk is 2.8% (95% CI 1.6 to 4.4), still above the level at which patients would like investigation (15). This suggests that clinicians should consider a cancer diagnosis in patients with a platelet count above these values. This could lead to earlier diagnosis, potentially at an earlier disease stage, if the patient is referred for further investigation sooner than they would have been had the raised platelet count not been recognized as a risk marker. A clinician receiving a high-normal platelet count should review why the test was done and what ongoing symptoms the patient is reporting. This finding is currently only a clue towards possible cancer. Until the findings are replicated in a much larger sample, and the specific cancers delineated in greater detail with data on stage at diagnosis, a blanket recommendation for investigation would be premature. Declaration Funding: The Policy Research Unit in Cancer Awareness, Screening and Early Diagnosis receives funding for a research programme from the Department of Health Policy Research Programme. It is a collaboration between researchers from seven institutions (Queen Mary University of London, University College London [UCL], King’s College London, London School of Hygiene and Tropical Medicine, Hull York Medical School, Durham University and the University of Exeter). OCU is supported by the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care South West Peninsula. The views expressed are those of the author(s) and not necessarily those of the National Health Service (NHS), the National Institute for Health Research (NIHR) or the Department of Health. Ethical approval: Ethical committee approval was given by the Independent Scientific Advisory Committee of the Clinical Practice Research Datalink: reference number 13-007. Conflict of interests: WH is an associate editor of Family Practice. The authors declare no other competing interests. References 1. Coleman MP, Forman D, Bryant H, et al.   Cancer survival in Australia, Canada, Denmark, Norway, Sweden, and the UK, 1995–2007 (the international cancer benchmarking partnership): an analysis of population-based cancer registry data. Lancet  2011; 377: 127– 38. Google Scholar CrossRef Search ADS PubMed  2. Walters S, Benitez-Majano S, Muller P, et al.   Is England closing the international gap in cancer survival? Br J Cancer  2015; 113: 848– 60. Google Scholar CrossRef Search ADS PubMed  3. Bailey SE, Ukoumunne OC, Shephard EA, Hamilton W. Clinical relevance of thrombocytosis in primary care: a prospective cohort study of cancer incidence using English electronic medical records and cancer registry data. Br J Gen Pract  2017; 67: e405– 13. Google Scholar CrossRef Search ADS PubMed  4. Biino G, Santimone I, Minelli C, et al.   Age- and sex-related variations in platelet count in Italy: a proposal of reference ranges based on 40987 subjects’ data. PLoS One  2013; 8: e54289. Google Scholar CrossRef Search ADS PubMed  5. NICE. Suspected cancer: recognition and referral [NG12]. 2015. http://www.nice.org.uk/guidance/NG12 (accessed on 5 September 2017). 6. Herrett E, Gallagher AM, Bhaskaran K, et al.   Data resource profile: Clinical Practice Research Datalink (CPRD). Int J Epidemiol  2015; 44: 827– 36. Google Scholar CrossRef Search ADS PubMed  7. StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC. 8. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol  2008; 61: 344– 9. Google Scholar CrossRef Search ADS PubMed  9. Bailey SE, Ukoumunne OC, Shephard E, Hamilton W. How useful is thrombocytosis in predicting an underlying cancer in primary care? A systematic review. Fam Pract  2017; 34: 4– 10. Google Scholar CrossRef Search ADS PubMed  10. Hamilton W, Lancashire R, Sharp D, Peters TJ, Cheng KK, Marshall T. The importance of anaemia in diagnosing colorectal cancer: a case—control study using electronic primary care records. Br J Cancer  2008; 98: 323– 7. Google Scholar CrossRef Search ADS PubMed  11. Shephard EA, Neal RD, Rose PW, Walter FM, Hamilton W. Symptoms of adult chronic and acute leukaemia before diagnosis: large primary care case-control studies using electronic records. Br J Gen Pract  2016; 66: e182– 8. Google Scholar CrossRef Search ADS PubMed  12. Shephard EA, Neal RD, Rose P, Walter FM, Litt EJ, Hamilton WT. Quantifying the risk of multiple myeloma from symptoms reported in primary care patients: a large case-control study using electronic records. Br J Gen Pract  2015; 65: e106– 13. Google Scholar CrossRef Search ADS PubMed  13. McGlynn EA, McDonald KM, Cassel CK. Measurement is essential for improving diagnosis and reducing diagnostic error: a report from the Institute of Medicine. JAMA  2015; 314: 2501– 2502. http://www. nationalacademies.org/hmd/Activities/Quality/Diagnostic ErrorHealthCare.aspx. 14. Hamilton F, Carroll R, Hamilton W, Salisbury C. The risk of cancer in primary care patients with hypercalcaemia: a cohort study using electronic records. Br J Cancer  2014; 111: 1410– 2. Google Scholar CrossRef Search ADS PubMed  15. Banks J, Hollinghurst S, Bigwood L, Peters TJ, Walter FM, Hamilton W. Preferences for cancer investigation: a vignette-based study of primary-care attendees. Lancet Oncol  2014; 15: 232– 40. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Family PracticeOxford University Press

Published: Apr 12, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off