Clinical Audit Report

Transcription

1 Clinical Audit Report The Lung Cancer Pathway from Initial Presentation to Healthcare Services until Diagnosis (Auckland & Lakes Regions) Identification of Barriers to the Early Diagnosis of People with Lung Cancer and Description of Best Practice Solutions Project Wendy Stevens, Graham Stevens, John Kolbe, Melissa Murray, Richard Hulme, John Cameron, Jeff Garrett, Chris Lewis, Denise Aitken, Matire Harwood December 00

2 Acknowledgements The clinical audit of the lung cancer pathway from presentation to healthcare services until diagnosis was conducted as part of the project Identification of barriers to the early diagnosis of people with lung cancer and description of best practice solutions. The project was funded by a Health Research Council and District Health Boards New Zealand (HRC_DHBNZ) Grant. The primary care audit was performed within the four participating primary care organisations, ProCare Networks Auckland and Manukau, and Total Healthcare Otara in Auckland, and Rotorua Area Primary Health Services and Pinnacle General Practice Network in the Lakes region. The secondary care audit was performed within Auckland District Health Board (DHB), Counties Manukau DHB and Lakes DHB. The study was designed and the data analysed by Dr Wendy Stevens. Primary care data were collected Dr Viv Tate (ProCare), Dr Richard Hulme (Total Healthcare Otara), Dr Rawiri Keenan (Pinnacle) and Dr Dean Tasker (Rotorua Area Primary Health Services). Secondary care data were collected by Dr Wendy Stevens and Ms Melissa Murray. Dr Sudha Purchuri (Auckland District Health Board) assisted with the reliability check of the data. Members of the Research Team involved in this component of the research included: Dr Wendy Stevens A/Prof Graham Stevens Prof John Kolbe A/Prof Jeffrey Garrett Dr Christopher Lewis Dr Denise Aitken Dr Richard Hulme Dr John Cameron Dr Matire Harwood Ms Melissa Murray Northern Cancer Network & University of Auckland Auckland DHB & University of Auckland Auckland DHB & University of Auckland Counties Manukau DHB Auckland DHB Lakes DHB Total Healthcare Otara ProCare Health Ltd Tamaki PHO University of Auckland The research team would like to thank all those within the DHBs and the primary care organisations who facilitated the conduct of the research. The research team is grateful to all those who contributed their time and expertise to the project.

3 Contents Page Executive summary. i. Background. Study Aims. Cases and Methods. Results. Patient and tumour characteristics. Initial presentation to health care services. Investigations and Referrals within Primary Care. Management within Secondary Care. Pathway Transit times. Retrospective Presentation Date & Raised Index of Suspicion 8.7 Comparison of Findings with the Previous Audit 8. Discussion... 0 Appendix A: Lung Cancer Clinical Pathway to Diagnosis.. Appendix B: Lung Cancer Timelines to Diagnosis. Appendix C: Cumulative plots showing the proportion of cases diagnosed over time Appendix D: Summary List of Main Findings. 0 References.. Abbreviations..

4 Executive summary The current study was undertaken to document the clinical pathway from initial presentation to healthcare services until diagnosis for people with lung cancer in the Auckland and Lakes regions. It built on the work of a previous audit of the secondary care management of lung cancer by some of the same investigators. It was part of a larger project Assessment of Barriers to the Early Diagnosis of Lung Cancer within Primary Care and Descriptions of Best Practice Solutions funded by a HRC_DHBNZ grant and conducted by the Northern Cancer Network in collaboration with the University of Auckland and other stakeholders. Study findings, together with those from other components of this project, will be used to inform recommendations to improve lung cancer care, regionally and nationally. The study involved all cases with primary lung cancer diagnosed in 008 in the New Zealand Cancer Registry (NZCR) that were also registered before diagnosis with one of the participating primary care organisations. These organisations covered 80% of the population within the Auckland, Counties Manukau and Lakes District Health Boards (DHBs). For the 7 eligible cases, data were collected from the primary and secondary care clinical records. The mean age of patients at diagnosis was 8yrs; 0% were female; % were of European ethnicity, % were Māori and % of Pacific ethnicity; 0% were never smokers; % of cases with NSCLC had metastatic disease at diagnosis and 7% of those with SCLC had extensive disease. Almost half (%) of all cases presented with a new or worsened cough; haemoptysis occurred in %, and symptoms suggestive of metastatic disease were reported in 0%. The duration of symptoms was poorly documented in both primary and secondary care records. Approximately threequarters (7%) of cases initially presented to primary care; % presented to the general practice with which they were registered and 0% presented to another practice. The GP ordered a chest Xray (CXR) on % of these cases. Few investigations other than a CXR, sputum cytology or blood tests were performed within primary care. Spirometry was rarely recorded. Following presentation to primary care, 0% of cases subsequently self presented to ED; 0% were referred for acute admission; % were referred to a respiratory specialist; % were referred to another specialist; % were already under secondary care and one case was admitted to a hospital outside the region. The other % presented directly to secondary care without first having seen a general practitioner (GP) about symptoms attributable to lung cancer. However 8% had consulted a GP about another problem within the preceding six months. Of the cases that presented directly to secondary care, most (% of all cases) selfpresented to ED, some (%) were already under secondary care when they presented (commonly with an incidental finding), and the remaining few (%) cases were transferred from outside the study region. The type and timeliness of management within secondary care varied considerably according to whether the patient was initially seen as an inpatient or outpatient. Inpatient/outpatient status was associated with tumour stage, with inpatients more commonly having advanced disease and outpatients more commonly having early stage disease. Overall, a chest CT scan was performed in % of cases; % being performed whilst an inpatient. A respiratory specialist saw 8% of cases. A bronchoscopy was performed in % of cases and a CT fine needle aspirate was performed in %. MDM discussion occurred for %, being more common for those seen by a respiratory specialist and those with NSCLC (rather than SCLC), but was less common for those with metastatic disease. i

5 Substantial differences in demographic characteristics existed between cases in the Auckland and Lakes regions, although few differences in presentation or management until diagnosis were evident. Cases from the Lakes region were more likely to be female, of Māori ethnicity and have a history of smoking. Tumour type varied, with adenocarcinoma being significantly more common in Auckland cases. Although fewer cases in Lakes had early stage nonsmall cell carcinoma, the difference in tumour stage was not statistically significant between the Lakes and Auckland regions. There was also no significant difference in either primary or secondary care management until diagnosis between Lakes and Auckland. Differences existed between ethnic groups in age at diagnosis, gender, socioeconomic status, comorbidity burden and smoking history. Māori cases tended to be younger, were more commonly female and smokers than the other ethnic groups. Differences in tumour characteristics also existed with squamous carcinoma being more common among Māori cases and adenocarcinoma being more common in European and Pacific cases. There was no significant difference in management until diagnosis between European and Māori cases. Pacific cases were more commonly referred for an acute admission by their GP; and this was accounted for by more advanced tumour stage. Although Pacific cases more commonly had metastatic disease, the difference was not statistically significant (most likely due to the small number of Pacific cases). Although did not attends were not common prediagnosis, the majority of cases that did not attend an appointment or investigation were of Māori or Pacific ethnicity. The median time from presentation to primary care until diagnosis was days. Large variations in times to diagnosis existed, with diagnosis taking longer than four months for % of cases that presented to primary care. The median time to diagnosis was considerably shorter (7 days) for cases that presented directly to secondary care. The type and timeliness of management within secondary care varied according to initial inpatient/outpatient status, with outpatients, who more commonly had early stage disease, having considerably longer times to diagnosis (8 vs days for those with metastatic disease). There was no significant difference in transit times between the Auckland and Lakes regions or between ethnic groups. In summary, patients took varied routes to diagnosis, with many patients presenting acutely to secondary care and only a minority being referred from a GP to a respiratory specialist. There were wide variations in transit times to diagnosis. Those with potentially curable disease took much longer to be diagnosed than those with advanced disease, especially if they were outpatients when first seen by secondary care. There were no major differences in the type or timeliness of management until diagnosis between the Auckland and Lakes regions or between ethnic groups. This is the first study in New Zealand to document the lung cancer clinical pathway from presentation until diagnosis. It has provided much baseline information which together with findings from other components of the research project will inform recommendations to improve lung cancer care. ii

6 The Lung Cancer Pathway in the Auckland and Lakes Regions from Initial Presentation to Healthcare Services until Diagnosis. Background Lung cancer is a major health problem in New Zealand (NZ) accounting for ~00 deaths annually. It is the leading cause of cancer deaths overall, and in Māori and Pacific peoples. Survival from lung cancer in NZ is poor compared with that in many other developed countries (year relative survival: 0.% in NZ vs 8% in Australia, USA and Canada); 7 and it is particularly poor for Māori (.%). Regional inequalities in lung cancer survival exist within NZ, with rural areas having significantly poorer survival outcomes than major urban areas (at least for nonmāori). For example, the fiveyear relative lung cancer survival rate in the Lakes region was.% compared with.%.7% in Auckland over the same period (0000). Poorer lung cancer survival in rural areas has been attributed, at least partially, to later presentation of people with lung cancer to healthcare services and later diagnosis. The high mortality from lung cancer worldwide is attributed to late presentation with advanced disease at diagnosis, which precludes curative resection. 7 Survival is strongly associated with tumour stage at diagnosis, and those diagnosed early in the 8 natural history of the disease have the greatest chance of long term survival. Whilst the fiveyear relative survival for patients with localised disease (stage I/II) is ~0%, it drops to % for those with locally advanced disease (stage III) and to % if distant metastases (stage IV) are present at the time of diagnosis. 0 Therefore earlier diagnosis of people with lung cancer, combined with timely anticancer treatment, has the potential to improve survival outcomes. A previous audit of secondary care management of lung cancer in the Auckland region suggested barriers to early diagnosis and anticancer treatment. Such barriers are believed to be greater for Māori and Pacific peoples, and those in rural areas. The current study was undertaken to document the clinical pathway of people with lung cancer from initial presentation to healthcare services until diagnosis. It builds on the work of the previous audit. The study was part of a larger project Assessment of Barriers to the Early Diagnosis of Lung Cancer within Primary Care and Descriptions of Best Practice Solutions funded by a HRC_DHBNZ grant and conducted by the Northern Cancer Network in collaboration with the University of Auckland and other stakeholders. Information on this project can be found on the Northern Cancer Network website: (Click on Research ). The findings from the current study together with those from other components of the research project will be used to inform recommendations for service change to improve lung cancer care both regionally and nationally.. Study Aims This retrospective study was undertaken to document the lung cancer pathway from presentation to diagnosis in the Auckland and Lakes regions in 008. It also sought to identify differences in the clinical pathway between the Auckland and Lakes regions and between the three major ethnic groups (being European, Māori and Pacific peoples).

7 . Cases and Methods Cases were potentially eligible if they were registered in the NZ Cancer Registry (NZCR) with a diagnosis of lung cancer (ICD0 C & C) in 008 i and also registered in the six month period leading up to diagnosis with one of the participating General Practice (GP) or Primary Health Organisations (PHOs) in the Auckland or Lakes regions. These organisations were ProCare Network Auckland, ProCare Network Manukau and Total Healthcare Otara (THO) in Auckland; and Rotorua Area Primary Health Services (RAPHS) and Pinnacle General Practice Network Taupo in the Lakes region. ii Potentially eligible cases were identified by cross referencing the NZCR listing with the databases of registered patients in the primary care organisations. Ethical approval for the study was obtained from the Northern X Regional Ethics Committee. Of the 8 iii potentially eligible cases, two cases did not have primary lung cancer, six cases were diagnosed elsewhere ( overseas and in another region) and three cases were diagnosed postmortem. Postmortem diagnoses were excluded as they did not contribute relevant information on the clinical pathway; two of these cases were never suspected clinically, and the third was known to have a respiratory nodule but was lost to followup after having declined investigations and appointments. Of the 7 eligible cases, five were unusual or complex cases. One initially presented overseas but had not been diagnosed and represented in NZ; two cases had presented some time previously but had repeatedly not attended appointments and represented in 008, and for another two cases initially diagnosed with lung cancer the diagnosis was subsequently revised. As these latter cases were initially diagnosed with primary lung cancer and transited the lung cancer clinical pathway to diagnosis within the Auckland and Lakes region, they were considered eligible for inclusion. Data were accessed from the primary and secondary care medical records of eligible cases. Data were collected from the primary care records by four GPs, one from each primary care organisation. Secondary care data were collected by two members of the research team. A reliability check was performed on 0% of all cases to ensure consistency and reliability in data collection. For most cases the primary care records were accessed electronically onsite at the individual general practices and some paper records were available. Centralised electronic access was possible for THO and RAPHS practices. Secondary care records were accessed electronically from Auckland District Health Board (ADHB) via a link with the other DHBs. The National Health Index (NHI), a unique personal identifier of all persons accessing the public health system in New Zealand, was used to link information from the different sources. The data collected included demographic information, comorbidity, tumour, presentation, and management details. Demographic data included the date of birth, i Although all cases were registered as being diagnosed in 008, % were actually diagnosed in other years: in 00; in 007 and in early 00. ii ProCare Network Auckland, ProCare Network Manukau, Total Healthcare Otara, Rotorua Area Primary Health Service and Pinnacle General Practice Network covered approximately 80% of the population within the Auckland, Counties Manukau and Lakes District Health Board. iii Eight of the 8 cases were registered at two of the participating primary care organisations. These 8 cases represented 80% of all (0) lung cancer cases on the NZCR listing from Auckland, Counties Manukau and Lakes DHBs and was consistent with the number (8) of lung cancer cases estimated on a proportional population basis.

8 gender, ethnicity (as recorded in the NZCR), New Zealand Deprivation Index 00 (NZDep), iv and smoking history (ever smoked, current smoker, packyears). Comorbidities were recorded and scored according to the internationally validated Charlson Comorbidity Index (CCI). Tumour details included histological type and clinicoradiological stage at diagnosis. Presenting symptoms and their duration, date of presentation, date the GP s index of suspicion was raised, details and dates of investigations, referrals, appointments, diagnosis, and did not attends (DNAs) were recorded. For cases with a histological diagnosis, the date of diagnosis was taken as the date of the first report confirming malignancy; for cases with a clinicoradiological diagnosis, the date was that of the first documentation of the presumptive diagnosis of lung cancer in the clinical records. In addition to the actual presentation date, the GPs collecting data from the primary care records documented when in retrospect the patient first presented with symptoms or signs that could be attributed to the lung cancer. Primary care records were not available for (%) cases. Four GP practices declined access to their medical records ( cases). In addition, two GPs had retired and their records were unavailable ( cases) and for two cases in a rest home their primary care records relating to the lung cancer presentation could not be located. Secondary care records were available for these cases. Some information, such as the date of the inpatient respiratory referral and/or consultation, was missing from the secondary care records for some cases. Data were double entered into an Excel database, cleaned, checked, merged and any discrepancies resolved. Data were imported into SPSS version (SPPS Inc., IL, 00) for analysis. The Chi Square test was used to assess associations between categorical variables and the Fischer Exact test was used for small numbers; Anova was used to assess differences in the means of numerical variables; and logistic regression was used to assess associations between factors and a dichotomous categorical variable. Log Rank test was used to assess differences between factors in cumulative proportion plots.. Results. Patient and tumour characteristics Patient and tumour characteristics of cases are shown in Table. The mean age of patients at diagnosis was 8yrs; 0% were female; % were of European ethnicity, % were Māori and % were of Pacific ethnicity; and 0% were never smokers. Smoking history was better recorded in the secondary care records than the primary care records which typically documented only whether the patient was a current smoker. Tumour stage was documented for % (8) of cases. Over half (%) of those with known stage had metastatic disease at diagnosis: % of cases with NSCLC had metastatic disease and 7% of those with SCLC had extensive disease at diagnosis. iv The NZDep is a censusbased, smallarea index of deprivation derived from the domiciliary address, which measures socioeconomic deprivation on a scale of to 0 (with 0 representing the most deprived level).

9 Table : Patient and tumour characteristics overall and in Auckland and Lakes Characteristic Age at diagnosis <0 yrs 0 yrs 707 yrs 80 yrs Mean Age yrs (SD) Median yrs (IQR) Range yrs Gender Male Female Ethnicity European Mäori Pacific peoples Asian Other Unstated DHB ADHB CMDHB LDHB Smoking status Current smoker Exsmoker Neversmoker Total Cases (7) Auckland Cases Lakes Cases (8) () N % N % N % yrs () yrs (; 7) yrs < < yrs () yrs (; 7) yrs < < 8 7 7yrs () 8yrs (; 7) 8 yrs p value <0.000 Unknown Ever Smoker Pack years Mean (SD) () () () NZDepDeciles (low deprivation) 78 0 (high deprivation) Comorbidities Respiratory COPD Other lung disease Diabetes Mellitus Cardiovascular Cerebrovascular Other Vascular Renal impairment Dementia Psychiatric disorder Alcohol abuse Another cancer in past CCI n/s for any 0.

10 Table continued Characteristic Tumour Type ClinicoRadiological Diagnosis NSCLC Carcinoid Tumours SCLC NSCLC Type Adeno Squamous NSC not further defined Other Unknown NSCLC Stage () Stage I Stage II Stage III IIIA IIIB Stage IV Unkown SCLC Stage () Limited Extensive Total Cases (7) Auckland Cases () Lakes Cases (8) N % N % N % () () 8 7 < 8 < () () < 8 7 < 8 0 () () p value Mean number of packyears smoked by smokers (nonsmokers excluded). NSCLC includes all tumours staged as NSCLC (ie histologically verified NSCLC, carcinoid tumours and clinicoradiological diagnoses) * Comparison of NSCLC vs SCLC using Fisher Exact Test (because of small numbers in a group) SD: Standard deviation; IQR: Interquartile range; NZDep: NZ Deprivation Index 00 p values relate to differences between Auckland and Lakes cases and were derived from i) Chi Square test for categorical variable (Fisher s Exact test was used if small numbers) and ii) Anova for difference between means for continuous variables * Regional differences As shown in Table, differences in demographic and tumour characteristics existed between the Auckland and Lakes cases. Cases in the Lakes region were more commonly female (% vs 7%; p 0.0), of Māori ethnicity (% vs %; p<0.000), and had a history of smoking (0.0). SCLC was more common in the Lakes region (although the number of SCLC cases was small). The type of NSCLC also varied (p 0.00), with adenocarcinoma being more common in Auckland (% vs %); however there was a high proportion of NSCLC not further defined in Lakes. Although fewer cases in Lakes had stage I/II NSCLC than in Auckland (% vs %), there was no statistically significant difference in tumour stage between the two regions (overall or for NSCLC or SCLC). Stage was unknown for % of cases (7% in the Lakes region and % in Auckland). Ethnic differences Substantial differences existed between ethnic groups in age at diagnosis (p 0.000); gender (p 0.0); socioeconomic status as measured by NZDep (p 0.0); comorbidity burden as measured by the CCI (p 0.0); and smoking history (p 0.00). The average age of European cases was 7yrs compared with yrs for Māori and yrs for Pacific cases. The majority of Māori cases were female (% compared with 7% of European and 8% of Pacific cases). Only % of European cases lived in the most socioeconomically deprived quintile, compared with % of Māori and 8% of Pacific cases. There was no significant difference in the CCI between European and Māori cases; although Māori cases were significantly more likely to have diabetes (0% Māori had diabetes compared with % European cases; p<0.000). There was

11 however a significant difference (p 0.00) in CCI between Pacific and other cases, with more Pacific cases having a significant comorbidity (7% Pacific cases had a CCI of 0 compared with 0% European and Māori cases). Pacific cases also more commonly had diabetes than European cases (% vs %; p 0.00). Māori were more commonly current smokers (% compared with % of Pacific cases and % of European cases) and fewer Māori were never smokers (% compared with % of European cases and % of Pacific cases). Māori cases more commonly had squamous carcinoma (% of Māori cases with NSCLC had squamous carcinoma compared with % European and % Pacific cases; p 0.0); whilst adenocarcinoma was more common in European (7%) and Pacific (%) cases than in Māori (%). However there was no significant difference between ethnic groups in the likelihood of NSCLC or SCLC. There was no significant difference (p 0.07) between the three major ethnic groups in tumour stage at diagnosis. Although Pacific cases more commonly had metastatic disease and less commonly had localised disease than European and Māori cases (Table ), the difference was not statistically significant, possibly due to the small number of Pacific cases () in the study. Table : NSCLC stage at diagnosis by ethnicity Ethnicity Localised (I/II) N Row% NSCLC Stage Locally Advanced (III) N Row% Metastatic (IV) N Row% Unknown N Row% European Māori 0 8 Pacific Asian Other 00 Unstated 00. Initial presentation to health care services No primary care records were available for (%) cases, although secondary care records were available. Another 78 (%) cases had no consultations relating to possible lung cancer documented in the clinical records at their registered practice, despite (8%) having had at least one consultation at the practice within the six months preceding diagnosis (Table ). Some () of those without a consultation during the six months preceding diagnosis had transferred practices (for four cases relevant information was available in the transfer documentation) whereas others had visited other practices. Table : Primary care consultations at the registered practice within the six months preceding diagnosis No available records No relevant consultations no consultation within months consultation within months Relevant consultation N % 78 () () 80 () (8)

12 Initial Presentation Overall, 7% (0) of all cases initially presented to primary care; % (80) presented to the general practice with which they were registered and 0% () presented to another practice. The remaining % () of cases presented directly to secondary care without first seeing a GP; % (7) selfpresented to emergency department (ED); % () were already under secondary care when they presented with symptoms () or had an incidental finding (); and four cases were transferred from a hospital outside the study region (Table ). Of the cases that presented to secondary care, % () had seen their GP within the preceding six months although they had not reported any symptom related to lung cancer; % (7) had not seen a GP within the preceding six months. Table : Initial presentation Initial Presentation All Cases (7) N % Primary Care Registered GP Practice Another GP Practice Secondary Care ED (selfreferral) Already under secondary care Transferred from another region 0 (80) () (7) () () 7 () (0) () () () There was no significant difference in the age, gender, ethnicity or tumour stage between those that initially presented to primary care and those that presented directly to secondary care. However, cases with a CCI (ie with multiple significant comorbidities) were less likely to have presented to primary care (odds ratio (OR) 0. (0.; 0.8; p 0.0) and more likely to already be under secondary care when they presented, commonly with an incidental finding. Presenting Symptoms As expected the most common presenting symptoms were cough, chest pain and shortness of breath. Almost half (%) of all cases experienced a new or worsened cough. Haemoptysis occurred in % of cases and symptoms suggestive of metastatic disease were present in 0% of cases. Almost onethird (%) of cases had substantial weight loss by the time they presented to secondary care. The frequency and duration of presenting symptoms is shown in Table. The duration of symptoms was poorly documented in both primary and secondary care records. Even though there was no significant difference in tumour stage between cases that initially presented to primary or secondary care, those presenting directly to secondary care were more likely to have symptoms suggestive of metastatic rather than locoregional symptoms (OR. (.0;.7); p 0.0). Almost onethird of initial secondary care presentations (%, cases) had an incidental finding on CXR or CT scan (Table ) and only one of these cases had any symptoms attributable to lung cancer. An additional cases presented to primary care with symptoms but were not suspected of having lung cancer until a CXR or chest CT scan was performed for another reason. 7

13 Table : Symptoms and their duration prior to presentation to primary and secondary care Symptom Primary Care Presentation Frequency N % (0) % with symptoms for mth mth Secondary Care Presentation of ALL cases (7) Frequency N % % with symptoms for mth mth Haemoptysis Unstated 0% Unstated % Cough 0 8 Unstated 0% Unstated % Shortness of breath Unstated % Unstated % Pneumonia 8 0 Chest pain Unstated % Unstated % Wheeze Unstated 0% Unstated % Dysphagia Unstated 0% Unstated 0% Hoarseness Unstated 7% Unstated 7% Palpable Lymph Node Superior Vena Cava Obstruction (SVCO) Weight Loss 8 Unstated 0% Unstated 0% Fever Fatigue 8 Nonspecific 7 Metastatic symptoms eg neurological symptoms; abdominal pain; mass etc Unstated 0% Unstated 0% 0% of cases with haemoptysis had it for less than week when they presented to primary care. Table : Initial symptom category overall and by initial presentation Characteristic Symptoms at Initial Presentation Locoregional symptoms Symptoms suggestive of metastatic disease Other or nonspecific symptoms Incidental finding Unkown Initial Presentation All Cases (7) Primary Care (0) Secondary Care () N % N % N %

14 Regional differences There was no significant difference between the Auckland and Lakes regions in whether patients initially presented to either primary or secondary care or in the type of presenting symptoms. Ethnic differences Presentation of the three major ethnic groups is shown in Table 7. Although Pacific cases presented more commonly with symptoms suggestive of metastatic disease and less commonly with an incidental finding than other ethnic groups, the difference was not statistically significant. Table 7: Ethnic differences in presentation Characteristic Initial Presenting Symptoms Locoregional symptoms Symptoms suggestive of metastatic disease Other or nonspecific symptoms Incidental finding Initial Presentation Primary care Secondary care European Māori 7 Pacific N % N % N % p value Investigations and Referrals within Primary Care GP management varied according to the type of presenting symptoms. A CXR was performed more commonly for those with locoregional symptoms than for those with other types of symptoms (0% vs 00%; p<0.000). However, referral to a specialist was more common in the presence of symptoms of metastatic disease (% vs %; p 0.00), as was referral for an acute admission (0% vs %); although the latter difference was not statistically significant. At the consultation when the patient initially presented with symptoms attributed to lung cancer, a CXR was ordered by the GP for 0% (8) cases; % were referred to a specialist and % were referred to the hospital for an acute admission (Table 8). A CXR was ordered for new respiratory symptoms (8%; 8 cases), worsening chronic symptoms (%; cases), unresolved pneumonia (7%; ), followup of a previous abnormal CXR (%; ), or for nonrespiratory symptoms (0%; 8) and the reason was unknown for five cases. The CXR was abnormal in % (80) cases, being suggestive of lung cancer in % ().

15 Table 8: Initial GP action overall and by symptom category Main Initial GP Action CXR Outpatient specialist referral Referral for acute admission Antibiotics Other symptomatic treatment Blood test Other Investigations Nil / review Unknown All Primary Care Presentations 0 Locoregional symptoms Metastatic Symptoms Other or nonspecific 0 N % N % N % N % At the following consultation, a CXR was ordered for 0 cases: of these had had a CXR ordered at the initial consultation but a subsequent CXR was ordered because of worsening respiratory symptoms in seven cases; unresolving pneumonia in one case and for nonrespiratory symptoms in three cases. All of these CXRs were abnormal with nine being suggestive of lung cancer. for the other cases, 0% () were ordered for new respiratory symptoms, 8% (8) for worsening chronic symptoms, % () for unresolving pneumonia, % () as followup of a previous abnormal CXR performed elsewhere, and % () because of nonrespiratory problems. All but one of these CXRs were abnormal; 7 being suggestive of lung cancer. In total, cases (% of cases that presented to primary care) had a CXR ordered by the GP; 8 initially, at the next consultation and subsequently. In addition, chest CT scans were ordered by a GP (in consultation with secondary care). Only % (8) of cases referred to hospital as an acute admission had a CXR prior to admission. Few investigations, other than sputum, blood tests or a CXR were performed within primary care. An abdominal ultrasound was ordered in five cases with abdominal symptoms, a mammogram in two cases, a barium swallow for dysphagia in one case, an MRI in one case with back pain, an ultrasound of the neck in one case with a supraclavicular node, and a fine needle biopsy in another case with a soft tissue mass. Only 7% (0) cases had spirometry recorded by the GP; although for % (7) cases secondary care spirometry results were available in the primary care records. Following presentation to primary care: 0% (0) of cases subsequently selfpresented to ED. The median time from initial presentation to primary care until presentation to ED was almost months (8 days; interquartile range IQR (8; ). Of these cases, the majority (80%) presented to the GP with locoregional symptoms. Half of the cases had moderate to severe COPD and all had a significant comorbidity ( had a CCI of ; had a CCI of ; and had a CCI ) which led to diagnostic difficulty. A CXR was performed on seven cases and was abnormal, being suggestive of lung cancer in three. Five cases were referred to a specialist however prior to the FSA they deteriorated and presented to ED. 0% () were referred to hospital for an acute admission. The majority (70%; ) presented with locoregional symptoms. A CXR was performed by the GP prior to the acute admission in 7% (); most () had an abnormal CXR with almost half 0

16 () being suggestive of lung cancer. Three cases had a chest CT scan performed prior to referral. Five cases were initially referred to a specialist but their condition deteriorated prior to the FSA and they were referred for an acute admission. The majority (7%; 7) of cases referred for an acute admission had presented to the GP with symptoms of lung cancer previously with the median time from initial presentation until referral being 7days (; 7). However % () were admitted acutely the first time they consulted the GP. % () were referred to a specialist: % (% of those referred to a specialist: 7) were referred to a respiratory specialist and % () were referred to another specialist. At Lakes DHB, patients were sometimes referred initially to general medicine, then the referrals were triaged to respiratory (this occurred for 0 out cases initially seen by respiratory at Lakes DHB.) Patients were referred to various specialties (general medicine (); neurology (); ORL (); general surgery (); gastro (); oncology (); and other). For 0 cases the GP rereferred the patient or contacted the specialist requesting an earlier appointment. In one case the original referral had been lost and in other cases the symptoms were worsening. Of the 7 cases referred to a respiratory specialist, 7% () had locoregional symptoms and 0% () had a CXR prior to referral, all of which were abnormal with most (0) being suggestive of lung cancer. A chest CT scan was ordered by the GP prefsa for 0 cases. The median time from initial presentation until referral was days (; 7). Five cases had particularly long times from presentation to referral of more than months (0 days): one had an initial normal CXR which reassured the GP; the other cases had significant comorbidities including severe COPD and other respiratory disease (such as TB) which led to diagnostic difficulty, as symptoms and abnormalities on the CXRs were attributed to these other diseases. Of the cases referred to another specialist % (8) had locoregional symptoms; % () had a CXR prior to referral, all of these were abnormal with most () being suggestive of lung cancer. The majority of cases with a suggestive CXR were referred to general medicine (: 0 of these were seen by a respiratory specialist at Lakes DHB); one was referred to elder care, three were referred directly to oncology; two were referred to ORL; three were referred to general surgery and one to neurology. Eight cases had a chest CT ordered by the GP pre FSA. The median time from initial presentation until referral was days (; 87). Six cases had particularly long times from presentation to referral of more than months (0 days): two cases did not attend the CXR resulting in delay; one case had severe COPD and recurrent symptoms were attributed to this; one case had dysphagia in the presence of a goitre and there was no obvious reason for the delay in two cases. % () were already being seen by secondary care when suspicion of lung cancer occurred even though they had consulted a GP. Four of these cases were under respiratory care (two for another respiratory problem; one for surveillance of a solitary pulmonary nodule and one for follow up postdischarge for recurrent pneumonia) and five cases were under another specialist for a concurrent problem when an abnormal CXR or CT scan was noted. one case that had consulted their GP was admitted whilst on holidays and subsequently transferred to a hospital in the study region. Regional differences Management within primary care did not differ significantly between the Auckland and Lakes regions, except that more cases were referred to a nonrespiratory specialist

17 in Lakes (p<0.000). However, this difference was due to different processes whereby cases with suspected lung cancer in the Lakes region were generally referred to general medicine and then triaged to a respiratory specialist. Ethnic differences The initial management of cases that first presented to primary care is shown by ethnicity in Table. Relatively more Pacific cases were referred for an acute admission (OR. (.; 7.); p 0.0) however the difference became not significant when tumour stage was adjusted for in the analysis. Half of these Pacific cases presented to the GP acutely and were immediately referred for an acute admission and the other half had consulted the GP previously regarding symptoms related to lung cancer. Table : Management within primary care for the major ethnic groups Main Initial GP Action CXR Outpatient specialist referral Referral for acute admission Antibiotics Other symptomatic treatment Blood test Other Investigations Nil / review Unknown European Māori Pacific N % N % N % 0 8 p value Because of small numbers GP action was grouped into categories with antibiotic, symptomatic treatment, blood tests, other investigation and nil/review forming one category Management within Secondary Care Cases entered secondary care by a variety of different routes as shown in Table 0. Just over half (%; ) of all cases were initially seen by secondary care as outpatients; and 7% (8) were seen as inpatients. i Of the cases seen acutely by secondary care, 8% (87) had previously seen a GP with symptoms attributable to lung cancer. Overall, 8% () of cases were seen by a respiratory specialist, % (0) had a bronchoscopy and % () had a CTFNA (Table ). In the majority () of cases a chest CT scan was performed prior to the bronchoscopy. However the bronchoscopy was nondiagnostic in % (7) of bronchoscopies (in % with a chest CT scan prior to the bronchoscopy and in % that had a bronchoscopy prior to the chest CT scan). A histological diagnosis of lung cancer was obtained in 8% (). The majority (77%) of cases with a clinicoradiologist diagnosis of lung cancer had metastatic disease (p<0.000). No patient had PET scan or CTPET scan prediagnosis. The management of cases within secondary care varied considerably according to initial inpatient/outpatient status (Table ). Whilst there was no significant difference in age, gender, NZDep, or CCI between cases initially seen as inpatients and those seen in outpatient clinics, differences existed in type of presenting symptoms and in tumour type and stage. Inpatients more commonly had symptoms i Cases seen in ED were included as inpatients as all (except one) were admitted from ED. The exception was admitted a few days later.

18 of metastatic disease (% vs 7%; p 0.00); SCLC (7% vs %; p 0.00); and those with NSCLC more commonly had metastatic disease (7% vs %; p<0.000) than outpatients Table 0: Entry route into secondary care overall and by initial presentation Characteristic Entry Route into 0 Care ED (selfreferral) GP referral for acute admission GP referral to respiratory GP referral to another specialist Already under respiratory Already under another specialist Initial Presentation All Cases (7) Primary Care (0) Secondary Care () N % N % N % 7 70 Other* case was referred by the district nurse and by the respiratory nurse (rather than by the GP). At Lakes DHB, cases were frequently referred to General Medicine and then the referrals were triaged so that cases with suspected lung cancer would be seen by a respiratory specialist. This occurred for 0 cases. * Transferred from outside the region Table : Management within secondary care overall and by in/outpatient status First specialist service seen Respiratory General Medicine General Surgery Other Seen by a respiratory specialist prior to diagnosis Chest CT Scan Inpatient CT scan GP ordered outpatient CT < All Cases 7 Outpatients Inpatients 8 N % N % N % p value p< Bronchoscopy Initial bronchoscopy negative nd bronchoscopy performed 7 CT FNA 7 <0.000 Number of other investigations 0 > Cytohistological diagnosis Clinicoradiological diagnosis Method of diagnosis Bronchoscopy 0 8 CTFNA Pleural aspirate 7 Biopsy of metastasis 8 7 Thoracotomy Other 7 7 MDM discussion <0.000 Number of investigations other than bronchoscopy or CTFNA eg CT head, CT abdo/pelvis; US abdo, MRI, biopsy of metastasis, bone scan, pleural aspirate etc % of cases with a cytohistological diagnosis

19 Outpatients Of the cases seen as outpatients, % (8) were seen privately. The majority (8%; 8) were first seen by respiratory and % () were seen by general medicine. In % (8) cases the GP was notified of the appointment date with the specialist. At some stage prior to diagnosis, 88% () of those initially seen as outpatients were seen by a respiratory specialist. Prediagnosis, five cases were seen by oncology but none were seen by palliative care. A chest CT was performed in % (), the majority (0) of these being performed as an outpatient with only a few () being ordered by the GP. A bronchoscopy was performed in % (7). Significantly more outpatients had a CTFNA (% vs %; p 0.000) than inpatients, presumably because of earlier stage disease in outpatients. A cytohistological diagnosis was obtained in 8% cases. Significantly more outpatients were discussed at an MDM (% vs % inpatients; p 0.00). Inpatients Of the 8 cases that were inpatients when first seen, % (7) selfpresented to ED, % (7) were referred for an acute admission (including that were initially referred to an outpatient clinic but their condition deterioration whilst waiting for the FSA) and % () were transferred from another hospital. Most cases were initially admitted under general medicine (70%; 0); % () under general surgery and 8% (0) under respiratory. The remaining % were admitted under other specialties. Prior to diagnosis, 8% (0) of these cases saw a respiratory specialist; 7% () were seen whilst an inpatient during the initial admission, % () were seen as an outpatient following the admission, and for two cases it was unknown where they were seen. During admission, eight cases were seen by oncology and four by palliative care. A chest CT scan was performed in % (), the majority () being performed as an inpatient. A bronchoscopy was performed in % (7) and a CTFNA in % (). A cytohistological diagnosis was obtained in 8% cases. Less than half (%) the inpatient cases were discussed at an MDM. MDM discussion Cases discussed at an MDM did not differ significantly in gender, ethnicity, CCI or NZDep from those not discussed. However, cases aged 80yrs or older were less likely to be discussed at an MDM than younger cases (OR 0.0 (0., 0.8; p 0.00). Cases with SCLC were less likely to be discussed than NSCLC cases (8% vs 0%; p 0.00); and cases with metastatic NSCLC were less likely to be discussed than cases with either locally advanced (7% vs 8%; p<0.000) or localised disease (7% vs 8%; p<0.00). Cases with locally advanced disease or localised disease were approximately five times more likely to be discussed than cases with metastatic disease (locally advanced: OR. (.,.); p<0.000) and localised disease: OR.7 (., 0.) ; p<0.000). Cases were also more likely to be discussed at an MDM if they had been seen by a respiratory specialist (% vs %; OR. (.;.); p<0.000). Those discussed at an MDM were eight times more likely to have a cytohistological diagnosis than those not discussed (OR 8.0 (.; 8.; p<0.000) and this association remained statistically significant after adjusting for tumour type and stage (p<0.000).

20 Regional differences There was no significant difference between Auckland and Lakes in the proportion of cases first seen as inpatients or outpatients, in the investigations performed or in the proportion of cases discussed at an MDM. Ethnic differences Pacific cases were significantly more likely than any other ethnic groups to be inpatients when first seen (7% Pacific vs 7% Māori and % European; p 0.0). However, this difference became not significant when tumour stage was adjusted for in the analysis. Although do not attends (DNAs) were not common prediagnosis (7%; 8 cases), the majority were of Māori or Pacific ethnicity: % Māori and 8% Pacific cases did not attend an appointment compared with % European cases (p 0.0). There was no significant difference between ethnic groups in the investigations performed, in the proportion that had a cytohistological diagnosis or in the proportion seen by a respiratory specialist or discussed at an MDM.. Pathway Transit times The lung cancer clinical pathway is shown in Figure (Appendix A) and the timelines to diagnosis are shown in Figure (Appendix B). Overall, the median time from presentation to primary care to diagnosis was days. However % of cases took longer than months ( days) to be diagnosed. For cases that initially presented to secondary care, the median time to diagnosis was only 7 days. Transit times within the lung cancer pathway varied considerably with a substantial proportion of cases having prolonged transit times whilst others had acceptable times (Table ). For cases with early stage (I/II) NSCLC the median time from primary care presentation until diagnosis was 00 days (7; ) compared with 8 days (; 0) for those with locally advanced disease (III) and only days (; 7) for those with metastatic disease (IV) (p<0.000). However there was no significant difference in times between cases with limited or extensive SCLC ( days vs 8 days). For cases with early stage (I/II) NSCLC the median time from when they were first seen by secondary care until diagnosis was days (; 7) compared with days (; 0) for cases with locally advanced disease (III) and days (; ) for cases with metastatic disease (IV). Cumulative plots showing the proportion of cases diagnosed by a certain time are presented in Figure (Appendix C). The type of presenting symptoms and inpatient or outpatient status significantly influenced transit times. Outpatient status, an incidental finding rather than symptoms, and early stage NSCLC were associated with longer times to diagnosis; whereas inpatient status, symptoms suggestive of metastatic disease and advanced stage disease were associated with shorter times to diagnosis. Haemoptysis was associated with significantly shorter times to diagnosis than other symptoms (p 0.0).

21 Table : Pathway transit times overall and by region Pathway Events Time from: Time from primary care presentation until CXR performed within 0 care ( cases; 7 missing dates) CXR requested until performed within 0 care ( cases; missing dates ) CXR within primary care until referral by GP ( cases; missing dates) Time from primary care presentation until GP referral received (7cases; missing dates) GP referral until FSA Receipt of referral until FSA ( cases; missing dates) Respiratory Referral to Resp FSA Outpatient Respiratory FSA ( cases; 0 with a missing date) Inpatient Respiratory FSA (0 cases; 8 missing referral dates) Primary care presentation to FSA ( cases; missing dates) Primary care presentation to Respiratory FSA (7 cases; 7 missing dates) Total Cases Auckland Cases Lakes Cases Median Time (Days) IQR Median Time (Days) IQR Median Time (Days) IQR (; ) (0; 8) 8 (; ) (0,) (0; ) (; ) (0,) (0; ) (0; ) (; 7) 8 (; ) (; ) 0 (same day) (8; ) (; 7) (; ) (0; 0) 0 (8; ) (; ) (; ) (0;0) 0 (8;) (0; ) (; ) (0; 0) (; 0) 0 (; 87) 8 (; 0) 0 (; 0) (; ) 8 (; 7) Chest CT scan requested to performed Outpatient CT ( cases; missing dates) Inpatient CT (0 cases; missing) (; ) (; ) (; 0) (; ) (; 8) (; ) FSA to Chest CT scan Outpatient CT ( cases; cases) Excluding cases if CT done prior to FSA Inpatient CT (0 cases; missing) (; 7) (; ) (, 7) (; ) (; ) (; ) Specimen Collection to Diagnosis () (; ) (; ) (0; ) FSA to Diagnosis Overall (7; missing) Outpatient when first seen by o care Inpatient when first seen by o care Respiratory FSA to Diagnosis ( cases; 8 missing dates) Overall (7; missing) Outpatient when first seen by Resp Inpatient when first seen by Resp Referral received to MDM FSA to MDM Primary Care Presentation to Diagnosis Overall (0cases; missing dates) Outpatient when first seen by o care Inpatient when first seen by o care Initial Presentation to Secondary care to Diagnosis ( cases; missing dates) ED or Acute Admission to Diagnosis (cases; missing dates) (8; ) (8; 8) (8; 7) (; 7) (7; ) (; ) (; 0) (0; ) (; ) (; ) (7; ) (8; 0) (; 8) (8; 0) (7; 8) (; ) (; ) (; ) (; 7) (; 0) (; ) (7; ) (; ) (; 7) (; ) (; ) (0; 0) (; ) (8; ) (; ) (; 7) (; 8) (7; 0) 7 (8; ) 7 (; ) (8; ) (8; 7) (8; 0) (; ) There was no significant difference in any of the above transit times between Auckland and Lakes.

22 Other factors associated with prolonged transit times were: Initial normal CXR: Although a normal initial CXR was uncommon (; %), the impact on transit times was considerable. For the four cases that presented to primary care the median time to diagnosis was days (; ) compared with days (; ) for all cases; and for the other five cases that initially presented to secondary care the median time to diagnosis was days (; 7) compared with 7 days (8; ) for all cases. Seven of the nine cases had metastatic NSCLC at diagnosis, one had stage IIB NSCLC and the other had unstated stage NSCLC. Initial bronchoscopy nondiagnostic: A nondiagnostic bronchoscopy increased the median time from FSA to diagnosis by 0 days ( days (7; 7) vs days (8; ) for all cases; p <0.000). Of the 7 cases that had a nondiagnostic initial bronchoscopy, 7 had a second bronchoscopy and had a CTFNA. In cases a histological diagnosis was attained and a clinicoradiological diagnosis was made in five cases. Five cases had SCLC ( limited and extensive) and had NSCLC (8 stage I/II; stage III; 8 stage IV and with unstated stage). Documented Delays In the clinical records there was evidence of delays due to both patient factors (%; cases) and system factors (0%; 8). The median time from presentation to primary care until diagnosis for those with a documented patient related delay was 77 days (; ) and 8 days (; ) for those with a documented system related delay compared with days (; ) for all cases. Patient related delays were caused by DNAs (7), initially declining investigations or specialist appointments (), the patient going on holidays (), or unspecified (). System related delays were suggested by the patient s condition deteriorating before the FSA (7); another referral being required as the previous referral had been lost, not acted upon or the appointment date was too far away (); or abnormal results not received or followed up (). For the 7 (%) cases that deteriorated prior to their FSA, five were referred for an acute admission; five selfpresented to ED and seven were rereferred urgently to a specialist. These cases had a prolonged median time from presentation to referral of 8 days (7; ) compared with days (; 7) for all cases. The median time from referral to being seen by secondary care was not increased ( days (; 8) vs days (0; 8) for all cases) although these cases were seen earlier than their booked appointment as their condition deteriorated. Regional & Ethnic differences There was no significant difference in transit times between the Auckland and Lakes regions or between ethnic groups.. Retrospective Presentation Date & Raised Index of Suspicion The GPs that collected data from the primary care records assessed the retrospective date of presentation (ie the date when the patient presented with symptoms or signs that in retrospect with the knowledge that the patient had lung cancer could be attributed to the lung cancer). This was not the date that a reasonable GP would necessarily be expected to suspect lung cancer, but rather the first indication of any symptoms or signs that could be attributed to the lung cancer. Of the 0 cases that initially presented to primary care there was sufficient information in the clinical records to determine a retrospective date of presentation for % () of cases. For 80% () of these cases, the retrospective date of presentation was the same as the actual date of presentation. For another 0%, the retrospective date was within three 7

23 months of the actual presentation date, and for the remaining 0% the retrospective date was between months to year earlier. Thus with the benefit of hind sight, 0% of lung cancer cases had an earlier presentation date within primary care. The date that GP s index of suspicion was raised either of lung cancer or another serious cause of the patient s symptoms was also documented. Of the 0 cases that presented to primary care, sufficient information was available for 8% (7) cases. For % (88) of these, the GP s index of suspicion was raised at the consultation when the patient initially presented. For % (7), suspicion was raised at a subsequent consultation most commonly because of persistent or worsening symptoms. For % () the GP never suspected lung cancer or another serious cause. The index of suspicion was raised within two weeks of presentation for 70%; within weeks to month for another 0%; months for 0%; and between months to year for the remaining cases. Delayed suspicion was commonly due to symptoms being attributed to other respiratory disease. The main findings of the current study are listed in Appendix D..7 Comparison of Findings with the Previous Audit As shown in Table, the findings of the secondary care component of the current study were similar to those of the previous audit of the secondary care management of people diagnosed with lung cancer in 00 in Auckland and Northland. In the current study there were relatively more females (0% vs %), more cases of Māori ethnicity (% vs 7%) and more cases from the most deprived quintile of the NZDep (0% vs %), due to the higher prevalence of these characteristics in the Lakes region, which was not part of the previous audit. In the current study, a higher proportion of cases entered secondary care acutely and fewer were already under secondary care; however this difference was due to definitional differences. As primary care data were not available in the previous audit, some cases that entered secondary care under a nonrespiratory specialist in the current study were classified in the previous audit as already being under secondary care. The investigations performed within secondary care and the methods of diagnosis were almost identical in the two studies. Considerably more cases were discussed at an MDM in the current study (% vs 8%). In both studies, those least likely to be discussed at an MDM were older cases, those with metastatic disease and those with SCLC. Timelines within secondary care were similar between the two studies with inpatients and those with metastatic disease having more rapid times to diagnosis than outpatients and those with earlier stage disease. It was disappointing that timeliness of diagnosis had not improved in the four years between the studies. 8

24 Table : Comparison of Findings with the Previous Audit Characteristic Age <0 yrs 0 yrs 707 yrs 80 yrs Gender Male Female Ethnicity European Mäori Pacific peoples Asian Other Unstated Tumour Type ClinicoRadiological NSCLC & Carcinoid SCLC Entry Route into 0 Care ED (selfreferral) or GP referral for acute admission GP referral to respiratory GP referral to another specialist Already under respiratory Already under another specialist Other Current Study (008) Auckland DHBs & Lakes DHB Secondary Care Audit (00) All Auckland DHBs & Northland DHB (7 cases) ( cases) % % First specialist service Respiratory Seen by respiratory 8 8 Histopathological Diag 8 8 Chest CT Scan Bronchoscopy CTFNA Method of diagnosis Bronchoscopy CTFNA Pleural aspirate Biopsy of metastasis Thoracotomy Other 0 0 < < 7 % with histological diagnosis * 8 % with histological diagnosis 7 MDM discussion 8 Median time to diagnosis (days) From presentation to 0 care (ie either receipt of referral or presentation at ED) All cases NSCLC Stage I/II Stage III Stage IV (0; ) 8 (; 7) (; ) (8; 0) (; ) (; ) (; ) (8; ) From ED/acute admission (8; 7) (; 8) Different definitions existed between the studies as primary care information was not available in the previous audit some cases referred to another specialist and some cases that presented either to ED or for an acute admission in the current study would have been classified as being already under secondary care in the previous audit. This would have accounted for most of the noted differences * 7% actually presented to ED and is the comparable figure here (% were initially referred to a specialist but presented to ED in the current study such cases were included in the ED presentations but in the previous audit they were included as specialist referrals )

25 . Discussion This is the first study in New Zealand to document the clinical pathway of lung cancer cases from presentation to healthcare services until diagnosis. The main purpose of the study was to provide baseline information to inform recommendations to improve lung cancer care. It built on work of a previous secondary care audit by some of the same investigators which documented the clinical pathway from secondary care presentation until the commencement of treatment. Lung cancer patients presented to healthcare services in a variety of ways and took varied routes to diagnosis. Threequarters (7%) of patients initially presented to primary care, most (%) to their own GP; and onequarter (%) presented directly to secondary care without first seeing a GP. These findings were similar to those in a large UK survey of cancer patients which indicated that 8% of lung cancer patients reported first seeing their GP. 7 The majority (%) of patients in the current study presented with locoregional symptoms, predominantly respiratory symptoms; whereas % had symptoms suggestive of metastatic disease and % had nonspecific symptoms. Similar to international figures 8, 8% had an incidental finding of lung cancer with no documented symptom of lung cancer. An incidental finding was made most commonly in patients being investigated by secondary care for another condition. The type of presenting symptoms influenced GP management and those with respiratory symptoms more commonly had a CXR and were referred to a specialist. As found in other studies 0 an abnormal CXR was an important trigger for referral to secondary care. Clear guidelines for ordering a CXR in primary care are therefore likely to facilitate timely referral and diagnosis. Although uncommon (%), a normal CXR lowered the index of suspicion resulting in considerable delay to referral and diagnosis. Only a minority (%) i of lung cancer patients followed the traditional route to diagnosis, being referred from a GP to a respiratory specialist. This is an important finding as most recommendations and guidelines focus solely on this traditional route, ignoring the other routes taken by the majority of lung cancer patients. UK studies also comment on this anomaly; although a substantially higher proportion of cases 0 (%) in the UK are referred to a respiratory specialist. A large proportion of patients presented to secondary care acutely, either being referred for an acute admission by the GP (%) or selfpresenting to ED (%). Almost half of those presenting acutely to secondary care (0% of all cases) had not presented to healthcare services with symptoms of lung cancer on a prior occasion. Although sudden development of symptoms without prior warning may have occurred in some patients, this suggests delayed presentation in a substantial number of patients. The nonspecific nature of lung cancer symptoms which are difficult to distinguish from those of other respiratory disease is considered an important factor in delaying diagnosis. Prompt recognition of lung cancer symptoms by both the patient and the GP is essential for early diagnosis and may be improved by increasing public awareness of lung cancer symptoms and ongoing GP education. 7 The findings of the current study relating to secondary care management were very similar to those of the previous secondary audit. Notable findings were that no i Even including those patients referred by their GP to general medicine who were actually first seen by respiratory, this figure increases to only 0% 0

26 patient had a PET scan; the bronchoscopy was negative in almost onethird of bronchoscopies, more commonly in those without a preceding chest CT scan. MDM discussion occurred in % of cases compared with 8% in the previous audit; this was most likely due to considerable efforts that occurred following the previous audit to increase MDM discussion. Several other recent changes have taken place in the Auckland region. A private PET scan facility has been established. DHBs within the Northern Cancer Network are now using a diagnostic algorithm which recommends a chest CT scan prior to bronchoscopy to facilitate case selection for bronchoscopy or CTFNA. A similar initiative was instigated in some parts of the UK to streamline the pathway to diagnosis. 8 The pathway within secondary care varied according to whether patients were inpatients (7%) or outpatients (%) when first seen by secondary care. Inpatients were more likely to have metastatic disease, symptoms suggestive of metastatic disease and to have SCLC. Inpatients had easier access to specialist assessment and diagnostic investigations with markedly reduced median times to diagnosis. The median time from primary care presentation to diagnosis was 8 days for outpatients compared with only days for inpatients. The median time from initial presentation to secondary care to diagnosis was only 7 days. Similar variations in times to diagnosis for different referral routes have been reported elsewhere. 0 Cases with advanced disease tended to be diagnosed more rapidly than those with early stage disease. Whilst it is considered important to diagnose and treat all patients promptly, it is those with early stage disease that are most likely to benefit from timely diagnosis. Although cure is possible for patients that present with early stage disease, tumour progression whilst awaiting diagnosis and treatment may render the tumour incurable. If improvements in lung cancer survival outcomes are to be achieved, interventions to improve the clinical pathway may need to be targeted specifically to those patients most likely to benefit from early diagnosis and treatment. Streamlined investigation of such patients (perhaps by day stay to avoid repeated outpatient visits) is required to facilitate earlier diagnosis of potentially curable cases. The overall median time from presentation to diagnosis in the current study ( days IQR: ; ) was considerably shorter than that reported in a comparable UK study ( days IQR: ; ). However survival outcomes for lung cancer in the UK are poorer than those in NZ (UK year relative survival rates 800: f 7.%; m % vs NZ 00: f %; m.%) and also this UK study involved an earlier time period (800), before UK pathway initiatives. 0 Considerable variation in transit times was evident in the current study with many cases experiencing prolonged times to diagnosis. Only half the patients had an outpatient FSA within the regional Cancer Network target of two weeks (median time: days (8; )); and only 7% were discussed at an MDM within the target weeks from referral receipt (median time: 8 days (; 0)). Although patient characteristics varied considerably between regions and ethnic groups, there was little difference in the pathway to diagnosis. Poorer lung cancer survival in rural areas has been attributed, at least partially, to later presentation. However in the current study, although fewer cases had early stage disease in the Lakes region compared with Auckland, the difference was not statistically significant. There was no significant difference in management up to diagnosis between the regions suggesting that other factors, such patient and tumour characteristics or treatment characteristics may be relevant. Although not statistically significant, Pacific cases presented later in the course of the disease than patients of other

27 ethnicity suggesting that barriers to access to healthcare services may be greater for Pacific peoples. Whether this related to health knowledge, financial, cultural or other barriers was unknown. Pacific cases were significantly more likely to be referred for an acute admission due to symptoms associated with more advanced tumour stage at presentation. No other difference in referral or investigation was apparent, consistent with findings from the previous audit which suggested that differences in management between ethnic groups occurred after, rather than before, diagnosis. This is the first study in New Zealand to document the clinical pathway from presentation to diagnosis. Although the number of cases in the study was small ( % of all lung cancer cases nationally), it is likely that study findings were representative of lung cancer cases in the Auckland and Lakes regions. The population covered by the study DHBs and primary care organisations comprised 80% of the total population of these regions. Also comparison with a previous audit of all lung cancer cases in Auckland indicated similar results for common components. However, although the study covered two different geographic regions of New Zealand with vastly different reported lung cancer survival outcomes, it is not known whether the findings are generalisable to all other regions, especially remote rural regions. Barriers to access to optimal care at any point along the clinical pathway have the potential to impact adversely on patient outcomes. International experience suggests that a coordinated regional approach to lung cancer care is necessary to improve access along the clinical pathway. Whilst all patients with suspected lung cancer should have a streamlined coordinated and timely journey along the lung cancer clinical pathway, improved survival outcomes may necessitate interventions that are targeted to specific patient groups, such as those with early stage disease, outpatients and Māori and Pacific patients.

28 Figure : Appendix A Lung Cancer Clinical Pathway to Diagnosis 00% 7% % CXR % Transferred from outside the region % GP % 7% 7% Chest CT 8% % Inpatients % ED (%) Acute Admissions (%) Outpatients Bronchoscopy % Other Investigations % Diagnosis % % Cytohistological Respiratory CTFNA Diagnosis (%) % 8% % Gen Med (8%) Already under Secondary Care % Chest CT 8% (% as inpatient) R e s p i r a t o r y 8% total seen by Resp N O n R e s p i r a t o r y (0%) % Clinicoradiological Diagnosis % MDM % % Respiratory (%) NonRespiratory (%)

29 Figure : Appendix B Lung Cancer Timelines to Diagnosis Outpatient Pathway(excludes those outpatients whose initial presentation occurred whilst they were already under secondary care) Presentation to Primary Care Referral FSA Chest CT Diagnosis Median Time & IQR (Days) days (; 0) 0days (8; 8) 8days (; 8) days (8; ) days (; 7) days (7; ) 8days (; ) Inpatient Pathway Presentation to Primary Care (; ) Referral 0 (0; ) ED Acute Admission Chest CT Diagnosis Median Time & IQR (Days) Transferred from outside region days (8; 7) days (; ) 8days (; ) days (7; )

30 Appendix C Figure : Cumulative plots showing the proportion of cases diagnosed over time i) Time to diagnosis for all cases that initially presented to primary care ii) Time to diagnosis by inpatient/outpatient status when first seen by secondary care Log Rank p<0.000 Log Rank p 0.007

31 iii) Time to diagnosis by presenting symptoms Log Rank p 0.00 (days) Log Rank p 0.0 (Incidental finding vs symptoms p 0.0) All cases (ie those that initially presented to either primary or secondary care)

32 iv) Time to diagnosis by NSCLC stage Log Rank p 0.00 Log Rank p<0.000 All cases (ie those that initially presented to either primary or secondary care) 7

33 v) Time from presentation to primary care until GP referral vi) Time to diagnosis from GP referral by entry route vii) Time to diagnosis by region Log Rank p 0.7 (days) vi) Time to diagnosis from GP referral by entry route (days) 8

34 vii) Time to diagnosis by region Log Rank p 0.7 Log Rank p 0. All cases (ie those that initially presented to either primary or secondary care)