Study selection

Database searching resulted in 5829 records. Following deletion of duplicates, 3556 records were screened by title and abstract and 3458 irrelevant records were excluded. The full texts of the remaining 98 records were obtained and screened. Of those, seven were included in this systematic review (Fig. 1).

Fig. 1: PRISMA flow diagram.

Study identification, screening, and selection process.

Study characteristics

Most of the studies were conducted in Denmark (n = 2) and England (n = 2) and were non-RCTs (n = 5). Sample size ranged widely between 7228 and 56,02029 participants and follow-up times varied from 330 to 37 months31. Five different interventions were used across the seven studies, including: (i) Combined public and HCP LC awareness campaigns;30,32 (ii) letters and continuing medical education (CME) meetings to educate general practitioners (GPs) about referral criteria for fast-track evaluation of patients with “reasonable suspicion” of LC (maximum 72 h waiting time for evaluation, which includes low dose computed tomography [LDCT]);33,34 (iii) a cancer fast-track programme (i.e., target of 30 days between well-founded suspicion of cancer by a GP and the start of treatment). Referrals to this programme can also originate from emergency departments or other clinical departments involved in routine monitoring or screening;29 (iv) the thoracic-trained advanced practice provider-led LC strategist programme to minimise diagnostic redundancy, streamline management decisions for indeterminate nodules, and expedite curative therapy. Once patients were referred from primary care to secondary care, an individual evaluation strategy was developed and followed for them;31 and (v) multi-disciplinary meetings, screensavers, and posters to reduce delay between initial suspicion of LC and measurement of serum calcium levels28. Of note, Hypercalcaemia is a serious complication of LC and is associated with poorer prognosis28. The full characteristics of the included studies are presented in Table 1.

Table 1 Study characteristics (n = 7).Quality appraisal and level of evidence assessment

All the included non-RCTs (n = 5) used appropriate data collection methods, outcome measures, and intervention administration. Outcome data were complete in all non-RCTs. Four non-RCTs had clear research questions. The study by Philips et al.31 did not have a clear aim statement, despite clearly stated hypotheses. Only one non-RCT reported that participants were representative of the target population33 and only one non-RCT reported that confounders were accounted for in the study design28. Both RCTs (n = 2) had clear research aims, performed randomisation appropriately, collected data in line with the research aims, had groups that were comparable at baseline, and reported on participant adherence to the assigned intervention30,34. However, the outcome assessor was not blinded in Gudlbrant et al.’s34 RCT.

Four studies scored 2+ on the Scottish Intercollegiate Guidelines Network27 level of evidence criteria, indicating well-conducted non-RCTs with a low risk of confounding or bias and a moderate probability that the relationship is causal29,31,32,33. Only one study scored 2++, indicating a well-conducted non-RCT with a low risk of confounding or bias and a moderate probability that the relationship is causal28. Both RCTs scored 1+ indicating well-conducted RCTs with a low risk of bias30,34. See Table 2 for quality and level of evidence assessment.

Table 2 Quality appraisal and level of evidence assessment.Synthesis of findings

Outcomes reported in the reviewed studies were categorised into four categories as follows: diagnostic intervals; referral and diagnosis patterns; stage distribution at diagnosis; and time interval from diagnosis to treatment.

Table 3 Data extraction and summary of findings (n = 7).Table 4 Visual representation of the key review findings.Diagnostic interval

Four studies aimed to reduce the diagnostic interval (i.e., the time from the first presentation with symptoms of LC until diagnosis35) using the LC strategist programme31, a community- and GP-targeted cancer awareness campaign30, information on LDCT and CME sessions34, and a multimodal quality improvement project28.

A retrospective review of the LC strategist programme found that time from suspicious findings on CT chest, chest X-ray, and to a lesser extent abdominal CT, to initiation of diagnostic workup of lung nodules for treatment or surveillance was significantly shorter with the programme in comparison to routine referral (3 vs 28 days respectively, p < 0.001)31. Following referral, the median time to workup was also significantly shorter with the programme in comparison to routine referral (1 vs 7 days respectively, p < 0.001)31.

In contrast, a concurrent community- and GP-targeted breast, prostate, colorectal, and LC awareness campaign found no statistically significant difference in the total diagnostic interval at community (i.e., public intervention) level (median total diagnostic interval = 114.5 days pre-test vs 114 days post-test, mean difference = 0.06, 95% confidence interval [CI] 0.39–0.5, p = 0.79) or at GP level (median total diagnostic interval = 115 days pre-test vs 125 days post-test, mean difference = 0.02, 95%CI 0.56–0.60, p = 0.45)30. Likewise, a study measuring the effect of an intervention to inform GPs about direct access to LDCT found no statistically significant difference in primary care interval (i.e., the time from the patient’s first symptomatic presentation in primary care until referral to secondary care35) between patients of GPs who received information about indications for LDCT (intervention group) (media n = 14 days, inter quartile intervals [IQI] = 4–53) and patients of GPs who did not receive this information (control group) (media n = 18 days, IQI = 5–69, Prevalence Ratio [PR] = 0.99, 95%CI 0.65–1.54, p = 0.455)34. Moreover, no statistically significant difference was found in the diagnostic interval between patients in the intervention group (median = 44 days, IQI = 17–83) and the control group (media n = 36 days, IQI = 17-112, PR = 0.8, 95%CI 0.5–1.27, p = 0.299). However, the primary care and diagnostic intervals in the intervention group were significantly shorter if the GP also participated in a 1-h small-group-based CME session (primary care interval median = 9 days [with CME] vs 37 days [without CME], p = 0.048; diagnostic interval median=23 days [with CME] vs 66 days [without CME], p = 0.008)34.

In their quality improvement project, Apthrop et al.28 used multidisciplinary meetings, screensavers, and posters encouraging secondary care physicians to order serum calcium levels in patients with a suspected diagnosis of LC. This project aimed to help reduce delay between initial suspicion of LC and ordering serum calcium levels during initial LC diagnostic workup in England. This project led to a statistically significant reduction in overall median time to ordering serum calcium levels in patients with a suspected diagnosis of LC, from 13 days pre-test (i.e., before the quality improvement project) to 7 days post-test (p = 0.001)28.

Referral and diagnosis patterns

Three studies reported on patterns of LC referral and diagnosis following implementation of a public awareness and GP training campaign32, a cancer fast-track programme29, and GP information and CME sessions on indications for LDCT33. Athey et al.32 delivered a public and GP LC awareness campaign in six English communities with high LC incidence served by 11 GP surgeries (intervention group). This campaign ran for six weeks and used a “push-pull” approach to “push” the public to seek help for symptoms of concern and encourage GPs to “pull” symptomatic individuals into appropriate services. Five other communities served by nine GP surgeries with similar demographics served as the control group. There was a 27% increase in the number of chest X-rays ordered in the intervention group compared to a 19% increase in the control group during the campaign and six months post-test. In comparison to pre-campaign, there was a sustained increase in chest X-rays requested in the intervention group (20% relative increase) in comparison to a 2% relative reduction in the control group (Incidence Rate Ratio [IRR] = 1.22, 95%CI 1.12–1.33, p = 0.001) at 12 months post-campaign. Moreover, LC diagnoses increased by 27% (relative increase) in the intervention group and fell by 10% (relative reduction) in the control group. However, this was not statistically significant (IRR = 1.42, 95%CI 0.83–2.44, p = 0.199)32.

In a study of a cancer fast-track programme in Catalonia, Prades et al.29 noted increased use of the programme over time, with 3336 patients with suspected LC referred via the programme in 2006, compared to 3841 patients in 2009. The proportion of all new LCs that were diagnosed through this programme fell from 60.2% (95%CI 59.8–63.4%) in 2006 to 53.2% (95%CI 51.5–54.9%) in 2009. GPs were the source of 60.6% of referrals to the fast-track programme in 2006 (95%CI 59–62.3%), falling to 41.4% (95%CI 39.7–42.9%) in 2009, demonstrating increased referrals from other sources such as hospital-based clinicians and services. The LC detection rate via the programme fell from 49.9% (95%CI 48.2–51.6%) in 2006 to 39.7% (95%CI 38.1–41.2%) in 2009. Prades et al.29 reported a statistically significant increase in GP compliance with cancer fast-track referral guidelines from 70.8% in 2006 (95%CI 69.1–72.1%) to 82.3% in 2009 (95%CI 81.1–83.5%).

In a cohort study nested in an RCT, Guldbrandt33 examined the use of a fast-track referral option for GPs for patients with suspected LC and the effect of GP education and awareness training on direct referral to LDCT. This education comprised a one-hour CME session and information about LDCT, including indications and Positive Predictive Values (PPV) for LC (i.e., the ratio of patients truly diagnosed as positive to all those who had positive test results). Results showed that, out of 648 patients directly referred to LDCT, absolute numbers of referrals were significantly higher (61%, 95%CI 54–66%) among GPs working in a clinic with one or more CME-participating GPs. However, the referral rate to LDCT via fast-track was 0.13 per 1000 adults per month (95%CI 0.09–0.19) for CME-participating GPs compared to 0.14 (95%CI 0.09–0.20) for non-participating GPs. The PPV for LC diagnosis due to referral to a fast-track LC pathway was 13.3% (95%CI 8.7–19.1%) for CME-participating GPs and 6.1% (95%CI 3–11%) for non-participating GPs (2.2 higher PPV). This was found to be statistically significant (p = 0.027)33.

Stage distribution at diagnosis

Three studies reported on LC stage at diagnosis following an intervention. Athey et al.32 examined LC stage at diagnosis following a “push-pull” LC awareness campaign, Guldbrandt et al.34 examined LC stage at diagnosis following an information programme and CME sessions on LDCT for GPs, and Philips et al.31 examined LC stage at diagnosis following the LC strategist programme. Athey et al.32 found no significant stage shift three months, six months, or one year following the LC “push-pull” awareness campaign. Similarly, Guldbrandt et al.34 reported a non-statistically significant difference in stage of LC at diagnosis between the intervention group (i.e., information and CME sessions on LDCT) and control group (p = 0.586 for advanced LC and p = 0.595 for localised LC). Philips et al.31 also found non-statistically significant difference in stage at diagnosis for the seven patients in the LC strategist programme and 33 routine referral patients who underwent surgery for LC. This was the only study to report on disease free survival and overall survival. It was found that six of the seven patients (85.7%) in the LC strategist programme cohort were found to have early-stage disease with a median time of 37 days from suspicious imaging to treatment31. In these six patients, with a median duration of follow up of 33 months, disease free survival and overall survival were 100% (i.e., no LC recurrence and no LC death). As for the routine referral group, 25 of 33 patients (75.7%) were found to have early-stage LC with a median time of 68 days from suspicious imaging to treatment. In these 25 patients, there were six recurrences (76% disease free survival) and no deaths (100% overall survival) over a median time of 35 months. The differences in survival rates between the LC strategist programme group and the routine referral group were not statistically significant31.

Time interval from diagnosis to treatment

The time from LC diagnosis to treatment was measured in two studies following two specialist programmes, namely the cancer fast-track programme29 and the LC strategist programme31. The latter study found that the time from suspicious imaging to definitive management plan was 14.5 days in the LC strategist programme and 46.5 days in routine referral (p < 0.001)31. It was also found that referral to the programme moved patients into low-risk nodule surveillance approximately one month earlier relative to routine referral (12.5 vs 39 days respectively, p < 0.001). Compared to routine referral, management through the programme also significantly reduced the median number of hospital trips (4 vs 6 respectively, p < 0.001), median number of clinicians seen (1.5 vs 2 respectively, p = 0.08), median number of diagnostic studies obtained (4 vs 5 respectively, p = 0.01), median time from suspicious radiological findings to diagnosis (30.5 vs 48 days respectively, p = 0.02), and median time from suspicious radiological findings to treatment (40.5 vs 68.5 days respectively, p = 0.02)31. Moreover, time from suspicious radiological findings to surgical resection was significantly shorter in patients managed through the programme in comparison to routine referral (38 vs 69 days respectively, p = 0.05). Among patients with early-stage non-small cell LC treated with radiation therapy, the LC strategist programme led to a substantial reduction in the time from suspicious radiological findings to initiation of treatment in comparison to routine referral (62.5 vs 122.5 days respectively, p = 0.08)31. Conversely, in the cancer fast-track programme, Prades et al.29 noted a variable trend in mean time from detection of suspected LC in primary care to start of initial treatment. The 30-day target was not achieved, with mean times of 30.8 days, 38.9 days, 32.25 days, and 36.7 days in 2006, 2007, 2008, and 2009 respectively. There was also an increase in the proportion of patients waiting between 30 and 45 days (23.7% in 2006 vs 26.1% in 2009) and over 45 days (13.6% in 2006 vs 22.6% in 2009) from the time of LC detection to initiation of treatment.