Lung cancer is the leading cause of cancer death in New Zealand with approximately 1,500 deaths per year.1 It has a significant impact due to the high rates of morbidity and mortality associated with the disease.2 Survival from lung cancer in New Zealand is poor with a five year survival of 9.5% for men and 11% for women.3,4 Internationally many countries achieve better survival outcomes than New Zealand, including Canada, Australia and Sweden where five-year survival is between 16–18%.5 Māori have a greater incidence of lung cancer, with Māori men having 2.0 times the incidence and Māori women having 3.4 times the incidence of Europeans/others.1,6 The age standardised mortality rate for Māori is 3.5 times that of non-Māori.7 One of the key reasons for the poor prognosis for newly diagnosed patients with lung cancer is that most patients present with advanced stage disease. Treatment is therefore generally palliative, with few patients being suitable for potentially curative treatment such as surgery or stereotactic ablative body radiotherapy (SABR).8,9 Another important influence in patient outcome is the tumour biology, for instance those with small cell lung cancer have a poorer prognosis.10,11 Non-small cell lung cancer (NSCLC) and those who are epidermal growth factor receptor (EGFR) positive have a better outcomes.

The New Zealand Cancer Registry (NZCR) is a population-based tumour registry, collecting and storing cancer incidence data. The new cancer registrations are mainly based on the pathology reports sent by reporting laboratories electronically. Other sources include discharge reports from publicly funded and private hospitals, death certificates and autopsy reports.12,13 Data collected in the NZCR includes demographic information (such as date of birth, gender and ethnicity) and tumour information (such as cancer site and extent of disease). The NZCR is the major source of “information on the incidence of, and mortality from, cancer” and “a basis for cancer survival studies and research programmes”.14 Its completeness and accuracy are vital for cancer control in New Zealand.

The completeness and accuracy of registrations in the NZCR have been reported to be diverse for different cancers.13,15–17 Approximately 88% of the breast cancer cases recorded in the NZCR have information on disease extent with a 94% accuracy rate in those with disease extent.13 For colon cancer 96% cases have disease extent with a 87% accuracy rate,15 but only 31% prostate cancer cases have disease extent with a 89% accuracy rate.16 An audit was conducted to assess the lung cancer data in NZCR using the data recorded in the Auckland and Northland regional databases in 2004.17 Of the 565 audit cases, 66 (12%) cases were not included in the NZCR, and one duplicate registration and 78 (14%) ineligible cases were identified in the NZCR. Only 58% of the lung cancer cases recorded in the NZCR have information on disease extent with a 79% accuracy rate in those with disease extent.17 The audit of the lung cancer data in the NZCR was conducted a decade ago, and improvement on data quality may have been achieved. An updated quality assessment on the lung cancer data is needed.

The Midland Respiratory Group has been collecting data on all newly diagnosed cases of lung cancer who are referred to their multidisciplinary review meetings onto an access database: Midland Lung Cancer register (MLCR). It has maintained a register of all patients seen since 2004 and the centre has relatively complete recording of cases for the Midland Cancer Region (Lakes, Waikato, Tairawhiti and Bay of Plenty District Health Boards (DHBs)) since 2007. These four DHBs serve a combined population of 700,000 and generate approximately 400 new cases of lung cancer a year. The region has 27% Māori population and of the over 2,000 cases on the register, 600 are Māori. The register includes data on date and source of presentation, results of investigations including CT and spirometry, date of diagnosis and pathological reporting. All patients are staged and mode of treatment is then recorded (radiotherapy, chemotherapy or surgery).

This study aims to report the characteristics of newly diagnosed lung cancer cases in the Midland Cancer Network region, and to examine the data accuracy of registrations in the NZCR in 2011–2015.

Methods

Data cleaning and verification

Experienced clinicians validated the lung cancer cases diagnosed in 2011–2015 in the MLCR by comparing the clinical records and the data recorded in the MLCR. Lung cancer (ICD code: C33, C34)18 diagnosed in the Lakes, Waikato, Tairawhiti and Bay of Plenty DHBs in 2011–2015 were extracted from the NZCR and the MLCR.

Registration duplications were removed from the two datasets. Then they were linked together by the National Health Index (NHI) number, a unique identifier assigned to every person who uses health and disability support services in New Zealand. We classified all the combined cases into three groups: 1) matching cancer cases in both datasets, 2) cancer cases identified in the MLCR only, 3) registrations identified in the NZCR only. For registrations identified in the NZCR only, clinical records were examined by clinicians to verify the lung cancer diagnosis: 1) lung cancer diagnosed in 2011–2015, 2) lung cancer diagnosed before 2011 or after 2015, 3) not lung cancer, 4) cannot confirm due to insufficient information.

Combined lung cancer dataset

To understand the characteristics of lung cancer cases in the Midland Cancer Network region, we created a combined dataset based on the NZCR data and the MLCR data. The combined lung cancer dataset included all cancer cases in the MLCR and the additional verified lung cancer cases diagnosed in 2011–2015 and recorded in the NZCR. We have categorised the cancer cell types into six groups: small cell, NSCLC, NSCLC-other, malignant carcinoid, others and unknown. The characteristics of these patients by ethnicity (Māori, Pacific and others) was explored, and the differences were examined by Chi-Square tests, with a p-value of less than 0.05 considered significant. The odds ratio of having small cell lung cancer compared to non-small cell lung cancer between Māori patients and non-Māori patients was estimated using the logistic regression model after adjustment for age, gender, smoking status (recorded in the MLCR), DHB and year of diagnosis.

Accuracy and completeness of the NZCR

The accuracy of demographic data of the verified lung cancer cases in the NZCR including cancer extent/stage, date of diagnosis, DHB, gender and ethnicity (Māori, Pacific, Asian, European and others) were compared with the MLCR data and clinical records. As demonstrated in the audit17 on lung cancer data in NZCR in 2004, the difference on date of diagnosis between the NZCR and regional dataset may be due to difference regarding the definition of date of diagnosis. The NZCR may collect the date of diagnosis from 1) date of operation or biopsy, 2) date of admission, 3) date of death if diagnosed at autopsy or 4) ‘approximate time between onset and death’ as reported by certifying doctor on the death certificate if the only notification of a cancer comes from the death certificate.16 However, a regional dataset may record the date of diagnosis from 1) date of issue of the first report confirming malignancy, 2) the date of final report suggesting invasive malignancy or 3) the first documentation of the diagnosis in the clinical records.17

The MLCR records the patient characteristics at the time of diagnosis including date of birth, gender, ethnicity and patient domicile. These data are taken from the patient’s hospital records at the time of diagnosis. Cancer stage is recorded after discussion at the multi-disciplinary meeting, using the Tumour Node Metastases (TNM) staging system.19 The NZCR applies the Surveillance Epidemiology and End Results (SEER) programme cancer staging definitions.20 The TNM classification was updated by the American Joint committee on Cancer. ‘T’ describes the extent of the primary tumour. ‘N’ describes the extent of regional lymph node metastasis. ‘M’ describes the occurrence of distant metastasis.19 The SEER staging definition was developed by the American National Cancer Institute. Extent at diagnosis in the NZCR is coded as B (limited to organ of origin), C (Extension to adjacent organs), D (Extension to regional lymph nodes), E (distant metastases) and F (unknown).20 In this study, stage IA and IB in the TNM system were considered to be extent B, stage IIA, IIB, IIIA and IIIB were comparable to extent C and D, and stage IV were extent E.17 All data analyses were performed in IBM SPSS statistics 25 (New York, US).

Results

Data cleaning and verification

We identified 2,125 lung cancer registrations in the NZCR, and 1,572 lung cancer registrations in the MLCR (Table 1) in 2011–2015. We found four duplicate lung cancer registrations in the NZCR (Figure 1). Of the cancer cases recorded in the MLCR, 1,482 (94.3%) lung cancer cases were also registered in the NZCR. There are 639 lung cancer registrations recorded in the NZCR that were not identified in the MLCR. After examining the clinical records of these 639 patients, 485 (75.9%) were confirmed to be diagnosed with lung cancer in 2011–2015, 22 (3.4%) were diagnosed with lung cancer before 2011 or after 2015, 72 (11.3%) did not have lung cancer, and 60 (9.4%) could not be verified due to lack of information in the clinical records. Of the 485 lung cancer cases recorded in the NZCR but not in the MLCR, 20 (4.1%) were stage I, 9 (1.9%) stage II, 39 (8.0%) stage III, 341 (70.3%) were stage IV and 76 (15.7%) with unknown stage. Of the 90 lung cancer cases recorded only in the MLCR, 43 (47.8%) had clinical diagnosis only without any pathology report. Twenty-six patients (28.9%) had stage I lung cancer, 7 (7.8%) stage II, 16 (17.8%) stage III, 36 (40.0%) stage IV and 5 (5.6%) with unknown cancer stage.

Table 1: Number of cancer cases by year of diagnosis and DHB.

‍

Figure 1: Flow chart of the Combined Midland Lung Cancer Dataset.

Combined Midland Lung Cancer Dataset

The combined dataset included 2,057 lung cancer cases (Table 2). The mean age of the cohort was 70 years old at diagnosis, and 52.7% of the patients were male. There were 656 (31.9%) Māori patients, 22 (1.1%) Pacific patients and 1,379 (67.0%) patients of other ethnicities. Waikato DHB has the largest number of lung cancer cases, followed by Bay of Plenty, Lakes and Tairawhiti DHB. Sixty-one percent of patients were diagnosed with metastatic cancer, while 11.0% of those who had been staged had stage I disease. Eighty percent of patients had NSCLC, while 16.9% had small cell lung cancer. Māori patients were more likely to be diagnosed at a younger age (p-value <0.001, mean age: 66 years for Māori and 72 years for non-Māori), and more likely to be female (p-value <0.001) than others. The difference of cancer stage at diagnosis between Māori and others was not significant (p-value =0.137). Almost 50%Māori patients were current smokers, another 47.8% were ex-smokers and only 2.6% had never smoked, compared to 27.4% current smokers, 61.9% ex-smokers and 10.7% never smoked for non-Māori patients (p-value <0.001). The proportions of missing data in smoking status, cancer stage and cell type between Māori and non-Māori were similar: 22.6% and 25.1% for smoking status; 3.5% and 4.6% for cancer stage; and 16.9% and 19.2% for cell type.

Table 2: Characteristics of lung cancer patients between Māori and Non-Māori in the combined dataset.

‍

After adjustment for age, gender, year of diagnosis and DHB (Table 3), the odds ratio of having small cell lung cancer for Māori patients compared to non-Māori patients was 1.55 (95% CI: 1.17–2.05). The odds ratio of having small cell lung cancer was 4.06 (95% CI:1.72–9.60) for current smokers and 2.68 (95% CI: 1.14–6.30) for ex-smokers compared to patients who never smoked. Patients in Bay of Plenty DHB were less likely to have small cell lung cancer compared to patients in Waikato DHB (odds ratio: 0.66, 95% CI: 0.48–0.92). Age, gender and year of diagnosis did not have a significant impact on the risk of being diagnosed with small cell lung cancer.

Table 3: Adjusted odds ratios of having small cell lung cancer compared to non-small cell lung cancer.

‍

Accuracy and completeness of the NZCR

The demographic data in the NZCR has high accuracy, with 99.2% accuracy for gender, 97.0% for ethnicity, 98.6% for DHB and 99.9% for date of birth among the 1,967 verified lung cancer cases in the NZCR (Table 4). For the date of cancer diagnosis, 21.5% of the cancer cases were on the same date and another 65.5% were within one month difference. However, the recording of cancer extent in the NZCR is poor, with 1,041 (55.3%) cancer cases recorded with correct cancer extent.

Table 4: Accuracy of characteristics of lung cancer patients in NZCR.

Discussion

Lung cancer in the Midland Cancer Network region is relatively common with over 400 new cases per year. There has been a 15% increase in the number of lung cancer cases between 2011 and 2015. This has implications on the capacity of the cancer services to manage the increasing workload. Sixty-one percent of patients present with stage IV disease at diagnosis, while only 17% had potentially curative stage I or II disease.

Early diagnosis is the key to improving the prognosis for lung cancer patients. Reasons for diagnostic delay in New Zealand are complex and multifactorial.21 Previous New Zealand research has demonstrated that patient delay is common and many are diagnosed in the emergency department, although most patients had seen a general practitioner (GP) before diagnosis. Possible interventions to improve early diagnosis include ‘community initiatives to educate and resource at-risk patients to seek help, supporting and resourcing primary care to increase timely referral and implementing strategies to reduce system complexity for GPs and patients, and the employment of care coordinators’.21

Māori patients were younger (mean age: 66 years) compared to non-Māori patients (mean age: 72 years). More Māori with lung cancer were women (55.3%), whereas in non-Māori the majority of cancers were in men (56.5%). This is likely due to differences in smoking status. In 2009, the smoking prevalence for Māori was 44% compared to 18% for non-Māori.22 Māori women have the highest smoking prevalence rate (48% in 2009).22,23 It is worth noting that 10.7% of non-Māori patients and 2.6% Māori patients diagnosed with lung cancer had never smoked.

We found no difference in the risk of diagnosing metastatic lung cancer between Māori and non-Māori patients, but Māori patients are more likely to have small cell lung cancer than others. Small cell lung cancer is strongly associated with smoking.24 This study showed that Māori patients are more likely to be smokers or ex-smokers than others. However, after adjustment for smoking status and other factors, Māori patients are still 1.55 times more likely to have small cell lung cancer than non-Māori patients. Reasons for this need further investigation. A 2010 audit conducted in the Auckland and Lakes region showed no significant difference in having small cell lung cancer between Māori and non-Māori patients.25 This study only included 57 Māori patients and may have been under powered to show the difference,25 while our study has 656 Māori patients and 126 of them have small cell lung cancer.

The NZCR and MLCR are both of great value in understanding the epidemiology of lung cancer in the Midland Cancer Network region. There was some under-reporting in both the NZCR (4%) and the MLCR (24%), and a 5% misrecording in the NZCR. The quality of demographic data of the lung cancer registrations in the NZCR is excellent, but the completeness and accuracy of lung cancer extent has not improved much since 2004: 35% with unknown cancer extent in this study and 42% in the 2004 audit; 81% of accuracy in this study and 77% in the 2004 audit among those with known cancer extent in the NZCR.17 Further efforts will be needed to improve the cancer extent recording that will be critical for lung cancer research and cancer control in New Zealand.

Around 87% (1712) of the registrations in the NZCR had a date of diagnosis within one month difference compared with the MLCR/clinical records. Though the definition of date of diagnosis in the MLCR is considered to be more reasonable by clinicians, it is more feasible for NZCR to collect the date of diagnosis based on their definition: 1) date of operation or biopsy, 2) date of admission, 3) date of death if diagnosed at autopsy or 4) ‘approximate time between onset and death’ as reported by certifying doctor on the death certificate.17 For the other 10% registrations whose date of diagnosis in the NZCR was more than one month different from that in the MLCR, the influence of this discrepancy may be substantial especially when these data are used for survival analysis.

The strengths of this study include that records were reviewed by experienced clinicians to ensure the accuracy of lung cancer registrations, cancer stage and cell type. Both sources of data were combined to explore the tumour characteristics between Māori and non-Māori patients. However, this study has some limitations. In the combined dataset, smoking status was not recorded in 24% of the lung cancer cases, cell type was unknown in 18% of the cases and cancer stage was unavailable in 4% of the cases.

Conclusion

The MLCR provides excellent clinical data on newly diagnosed lung cancer cases. However, there is some under-reporting compared with the NZCR. Combining the two sources of data gives a more complete picture of the incidence and tumour characteristics of lung cancer in our region. Our combined dataset suggests that Māori patients are more likely to have small cell lung cancer than non-Māori patients.

‍