Health Care Utilisation and Health Needs of People with Severe COPD in the Southern Region of New Zealand: A Retrospective Case Note Review

Abstract

We examined health care utilisation and needs of people with severe COPD in the low-population-density setting of the Southern Region of New Zealand (NZ). We undertook a retrospective case note review of patients with COPD coded as having an emergency department attendance and/or admission with at least one acute exacerbation during 2015 to hospitals in the Southern Region of NZ. Data were collected and analysed from 340 case notes pertaining to: demographics, hospital admissions, outpatient contacts, pulmonary rehabilitation, advance care planning and comorbidities. Geometric mean (95%CI) length of stay for hospital admissions in urban and rural hospitals was 3.0 (2.7-3.4) and 4.0 (2.9-5.4) days respectively. More patients were from areas of higher deprivation but median hospital length of stay for patients from the least deprived areas was 2.0 days longer than others (p = 0.04). There was a median of 4 (range 0-16) comorbidities and 10 medications (range 0-25) per person. Of 169 cases where data was available, 26 (15%) were offered, 17 (10%) declined, and 5 (3%) completed, pulmonary rehabilitation at or in the year prior to the index admission. Patients were less likely to be offered pulmonary rehabilitation if they lived >20km away from the hospital where it took place (odds ratio of 0.12 for those living further away [95%CI 0.02-0.93, p = 0.04]). There were deficits in care: provision and uptake of non-pharmacological interventions was suboptimal and unevenly distributed across the region. Further research is needed to develop and evaluate strategies for delivering non-pharmacological interventions in this setting.

Keywords:

• Chronic obstructive pulmonary disease
• health systems service delivery
• pulmonary rehabilitation

Introduction

Chronic obstructive pulmonary disease (COPD) is a major chronic disease in New Zealand (NZ) with a high prevalence (14% of adults aged over 40), morbidity and mortality (the fourth leading cause of death in NZ in 2011) [1,2]. People with severe COPD (as assessed by symptoms, exacerbations and airflow obstruction) [3] account for the majority of COPD morbidity and mortality [4] and are commonly comorbid [5]. They experience breathlessness, fatigue, and impaired quality of life [3]. COPD has a higher prevalence in Māori and is associated with higher rates of hospitalisation in Māori and Pacific Peoples, the elderly, lower socioeconomic groups and those living in rural areas [6].

COPD places a large burden on the NZ healthcare system: the cost is estimated at $NZ 5.6 billion with $NZ 484 million in direct healthcare costs [1]. The main direct cost incurred is hospital admission secondary to an infective exacerbation but the burden on primary and community care and carers is also considerable [6].

There is good evidence that the burden of COPD can be lessened by the appropriate use of a range of non-pharmacological and pharmacological interventions. For example, pulmonary rehabilitation reduces mortality, morbidity and hospitalisations, and clinical guidelines recommend it is offered to all people with stable COPD [7]. Another example is that of integrated disease management programmes that span primary and secondary care and involve medical, nursing, allied health teams and social services. These may address issues such as smoking cessation, pulmonary rehabilitation, self-management and optimal medication adherence. This integrated approach to treatment is recognised as improving care and reducing hospitalisations [8]. Better integration of chronic disease management across primary and secondary care is a NZ health priority [9]. There is, however, evidence that people with severe COPD receive suboptimal care in NZ, most notably in relation to provision and uptake of pulmonary rehabilitation [10].

NZ’s Southern Region covers a population of 300,000 across a land area of 62,000 km², with 42% of the population living outside the two main urban centres of Dunedin and Invercargill. The population density of 5 people per km² compares with a NZ average of 18 and an average of 38 people per km in the member countries of the Organisation for Economic Co-operation and Development [2,11]. In the NZ Southern Region, the District Health Board (DHB) and Primary Health Organisation (PHO) are working together to improve service integration [12] and have recently developed a strategic health services plan [13] in which better chronic disease management is a key priority. COPD has been chosen as an exemplar condition due to its high prevalence and morbidity, the need to deliver interventions such as pulmonary rehabilitation to a dispersed rural population and the need to develop an integrated model of care that better meets the needs of patients with severe disease and their carers.

We aimed to determine health care utilisation and health care and social support needs of people with severe COPD. As part of a larger piece of work, which included a qualitative study of the views of patients and health care professionals [14], we undertook a retrospective case notes review of patients admitted to hospital across the Southern Region with exacerbations of COPD during 2015.

Methods

Study population

Patients were required to meet the following inclusion criteria:

• Seen in the emergency department and/or admitted as inpatients to any hospital across the Southern DHB (Dunedin Public Hospital, Southland Hospital, Invercargill; Lakes District Hospital, Queenstown; Oamaru Hospital; Clutha Health First Hospital, Balclutha; Maniototo Hospital, Ranfurly; Gore Hospital; Dunstan Hospital, Clyde).

• Coded as having at least one visit to hospital for an acute exacerbation of COPD over the 12-month period from 1 January to 31 December 2015.

• Diagnosis of COPD previously confirmed on post-bronchodilator spirometry (forced expiratory volume in one second/forced vital capacity <70%).

For those patients who had more than one hospital visit for an acute exacerbation of COPD during 2015, the visit closest to 31 December 2015 was taken as the index admission.

Data collection

We conducted a retrospective case note review and collected data pertaining to:

• Demographics (age, sex, ethnicity, rurality and social deprivation)

• Disease-specific data (most recent spirometry, smoking status)

• Details of hospital admission(s) (length of stay, number of admissions in prior 12 months)

• Outpatient contacts over the prior 12 and subsequent 3 months (outpatient appointments, allied health contacts)

• Invitation to and receipt of (hospital-based) pulmonary rehabilitation

• Evidence of discussion of advance care planning and resuscitation preferences

• Clinical comorbidities and medications prescribed

Data were collected using a secure web application for online databases (REDCap™) available via the University of Otago. An online form was created within this web application for data entry for each patient. Information from a patient’s electronic and paper notes was entered, eventually culiminating in a dataset. The full dataset was then de-identified for analysis.

Data analysis

Where at least 10 events and 10 non-events per degree of freedom were available, univariable binary logistic regression was used to model the odds of being offered pulmonary rehabilitation. Otherwise, and for all models of declining this rehabilitation, univariable exact logistic regression was used with conditional probability tests used to evaluate statistical significance for predictors.

For the numbers of hospital admissions that were COPD-related, univariable Poisson regression was used with univariable negative binomial regression used instead where a likelihood ratio test indicated over-dispersion. For lengths of stay (COPD and non-COPD), linear regression was initially used and histograms of residuals and scatterplots of residuals against fitted values used to assess model assumptions. Where these assumptions were not satisfied, univariable quantile regression was used instead to model medians.

Analyses were performed using Stata 15.1 and two-sided p < 0.05 was considered statistically significant in all cases. No adjustments were made for multiplicity as this was planned as an exploratory study and marginally statistically significant results should be interpreted with caution. No formal treatment was used for missing data and all available data was used for each analysis, consequently sample sizes vary between models with this indicated alongside all results in tables.

Ethical approval

Ethical approval was obtained from the Northern B Health and Disability Ethics Committee (HDEC) (16/NTB/11).

Results

540 patients who had at least one visit to hospital coded as an acute exacerbation of COPD were identified across the Southern DHB in 2015. Of these, 340 patients had COPD confirmed by spirometry at some time prior to their index admission and thus met criteria for inclusion in the study. Patient demographics are presented in Table 1. Mean age at index admission was 63.4 years in Māori and 72.5 years in NZ Europeans (the numbers of other ethnic groups were too small for comparison).

Table 1. Demographics of 340 patients in the Southern District with spirometry-confirmed COPD and at least one hospital visit for an exacerbation in 2015.

		Dunedin	Southland	Other	Total
n		169	123	48	340
Female		71 (42)	55 (45)	21 (44)	147 (43)
Age (mean (SD))		72.0 (10.4)	71.0 (10.1)	72.9 (10.6)	71.8 (10.3)
Ethnicity	European	153 (93)	101 (83)	40 (95)	294 (90)
	Māori	9 (5)	17 (14)	2 (5)	28 (9)
	Pacific Peoples	3 (2)	1 (1)	0 (0)	4 (1)
	Other	0	2 (2)	0 (0)	2 (1)
	Total	165 (100)	121 (100)	42 (100)	328 (100)
Number of comorbidities	0-2	31 (21)	23 (23)	9 (24)	63 (22)
	3-5	68 (45)	43 (43)	14 (37)	125 (44)
	6-8	33 (22)	14 (14)	10 (26)	57 (20)
	>9	18 (12)	19 (19)	5 (13)	42 (15)
	Total	150 (100)	99 (100)	38 (100)	287 (100)
NZ deprivation quintile*	1	24 (15)	7 (6)	4 (9)	35 (11)
	2	31 (19)	13 (11)	18 (38)	62 (19)
	3	35 (21)	30 (25)	14 (30)	79 (24)
	4	36 (22)	39 (32)	7 (15)	82 (25)
	5	38 (23)	32 (26)	4 (9)	74 (22)
	Total	164 (100)	121 (100)	47 (100)	332 (100)
Urban area^#	Main	135 (82)	83 (69)	0 (0)	218 (66)
	Minor	9 (5)	19 (16)	26 (55)	54 (16)
	Secondary	2 (1)	0 (0)	9 (19)	11 (3)
Rural area^#	Rural centre	8 (5)	13 (11)	3 (6)	24 (7)
	Other rural	10 (6)	6 (5)	9 (19)	25 (8)
	Total	164 (100)	121 (100)	47 (100)	332 (100)
Road distance from home to hospital (km)	0-10	91 (55)	81 (67)	27 (56)	199 (60)
	10-20	46 (28)	3 (2)	6 (13)	55 (16)
	20-50	5 (3)	15 (12)	10 (21)	30 (9)
	50-100	12 (7)	18 (15)	4 (8)	34 (10)
	>100	11 (7)	4 (3)	1 (2)	16 (5)
	Total	165 (100)	121 (100)	48 (100)	334 (100)

Data given as no. of patients (%) except age (SD) (total data for age: n = 334). *NZ deprivation quintiles: level of deprivation increases from 1 to 5. ^#Defined according to Stats NZ urban/rural profile.

Utilisation of acute hospital services

Of the 340 index admissions, 52 (15%) were visits to an emergency department only, while 288 (85%) involved hospital admissions (see Table 2). The geometric mean (95% CI) length of stay for hospital admissions in urban and rural hospitals was 3.0 (2.7 to 3.4) days and 4.0 (2.9 to 5.4) days respectively. Arithmetic mean length of stay for hospital admissions was 4.4 days. It was similar for 136 admissions to Dunedin and 105 admissions to Invercargill (4.2 and 4.1 days respectively), while it was 5.9 days for the 41 admissions to rural hospitals. Median length of stay for patients from the least deprived areas (NZDep deciles 1 to 3) was 2.0 days longer than other patients (NZDep deciles 4 to 10) (p = 0.04).

Table 2. Utilisation of inpatient services at Dunedin, Southland and other hospitals in the Southern District.

		Dunedin	Southland	Other	Total
Index admission type	ED only	33	18	1	52
	Inpatient	136	105	47	288
	Total	169	123	48	340
Index admission LOS (inpatient admissions only)		4.2	4.1	5.9	4.4
Mean total LOS (days) in 12 months prior to (and excluding) index admission (n = 337)		9.2	9.6	6.6	9.0
Mean no. of hospital admissions in 12 months prior to (and excluding) index admission (n = 337)		2.1	1.6	1.3	1.8
Mean total COPD LOS (days) in 12 months prior to (and excluding) index admission (n = 337)		4.7	5.7	3.1	4.8
Mean no. of hospital admissions for COPD in 12 months prior to (and excluding) index admission (n = 337)		1.1	1.1	0.7	1.0
Repeat admissions within 30 days of index admission (%) (n = 338)	All causes	33 (20)	19 (15)	8 (17)	60 (18)
	COPD	5 (3)	0 (0)	0 (0)	5 (1)
	Other resp	11 (7)	12 (10)	4 (8)	27 (8)
	Other cause	17 (10)	7 (6)	4 (8)	28 (8)

Abbreviations: ED = emergency department; LOS = length of stay; resp = respiratory.

Patients spent a mean of 1.8 admissions and 9.0 days in hospital in the 12 months prior to the index admission. Of these, a mean of 1.1 admissions and 4.8 days were for acute exacerbations of COPD. We observed no association between the number of COPD admissions in the 12 months prior to admission and age, sex, ethnicity, admission location, distance between home and hospital, rurality or social deprivation although those with comorbidities had nearly twice the number of admissions (incident rate ratio 1.99, 95% CI 1.05–3.76, p = 0.035 from negative binomial regression).

60 of 340 patients (18%) had a repeat admission within 30 days and, of these, 10 (3%) patients had two admissions within 30 days. 32 patients (10%) had a repeat admission for an acute exacerbation of COPD or other acute respiratory illness.

Utilisation of outpatient and allied health services

Table 3 shows utilisation of outpatient and allied health services in the 12 months prior to and 3 months after the index admission. Patients attending hospitals other than Dunedin or Southland for their index admission had fewer contacts with nursing and allied health staff, and differences in contacts with nursing staff accounted for most of this difference (a mean of 6.6 contacts in 15 months compared with 8.5 in Dunedin Hospital and 13.2 in Southland Hospital).

Table 3. Utilisation of outpatient and allied health services at Dunedin, Southland and other hospitals in the Southern District.

	Dunedin	Southland	Other	Total
All medical specialists	5.0	5.4	3.5	4.9
Respiratory specialists	0.7	1.0	0.8	0.8
Total allied health contacts	13.8	16.6	11.8	14.5
Respiratory specialist nurse	2.4	4.9	2.6	3.3
Palliative care nurse	1.1	0.1	0.0	0.6
District nurse	5.1	8.2	4.0	6.0
Physiotherapist	1.1	0.4	1.9	1.0
Occupational therapist	0.7	0.4	1.2	0.7
Health support worker	0.2	0.0	0.3	0.1
Social worker	0.1	0.1	0.3	0.1
Cultural support worker	0.4	0.9	0.0	0.6
Others	2.9	1.5	1.4	2.2

Values are mean number of contacts in the 12 months prior to and 3 months after the index admission (n = 340).

Health service utilisation by Māori and NZ Europeans

28 and 287 records were available for comparison of health service utilisation by Māori and NZ Europeans respectively (the numbers of other ethnic groups were too small for inclusion). A similar pattern of service use was seen between the two (Table 4).

Table 4. Comparison of health service utilisation by Māori and NZ Europeans in the study cohort.

	Ethnicity
	Māori	NZ European
Index admission LOS (days) (inpatient admissions only) (n = 21 for Māori, n = 246 for NZ European)	4.0	4.5
Total LOS (days) in 12 months prior to (and excluding) index admission (n = 28 for Māori, n = 285 for NZ European)	9.8	9.2
Total COPD LOS (days) in 12 months prior to (and excluding) index admission (n = 28 for Māori, n = 285 for NZ European)	4.7	5.1
No. of hospital admissions in 12 months prior to (and excluding) index admission (n = 28 for Māori, n = 285 for NZ European)	1.9	1.9
No. of hospital admissions for COPD in 12 months prior to (and excluding) index admission (n = 28 for Māori, n = 285 for NZ European)	1.0	1.1
No. of contacts with all medical specialists	6.1	4.8
No. of contacts with respiratory specialists	1.1	0.8
No. of contacts with respiratory specialist nurse	9.1	6.0
No. of contacts with palliative care nurse	1.0	5.4
No. of contacts with district nurse	21.0	21.8

Values are means. n = 28 for Māori and n = 287 for NZ Europeans unless otherwise stated.

Abbreviation: LOS = length of stay; NZ = NZ.

Comorbidities

Of 340 records, 287 listed the patient’s comorbidities at the index admission. There was a median of 4 comorbidities per person (range 0-16), and Table 1 shows further detail on the numbers of patients with grouped numbers of comorbidities. Table 5 shows the numbers and proportions of Māori and NZ European men and women in the study cohort with cardiovascular and other relevant comorbidities.

Table 5. Numbers of Māori and NZ European men and women in the study cohort with cardiovascular and other relevant comorbidities (percentages are given in brackets).

	Māori						NZ European
	Male		Female		Total		Male		Female		Total
	(n = 13)		(n = 8)		(n = 21)		(n = 134)		(n = 111)		(n = 245)
MI	0	(0)	0	(0)	0	(0)	17	(13)	5	(5)	22	(9)
Heart failure	2	(15)	1	(13)	3	(14)	10	(7)	10	(9)	20	(8)
Hypertension	9	(69)	3	(38)	12	(57)	51	(38)	47	(42)	98	(40)
IHD	6	(46)	0	(0)	6	(29)	38	(28)	21	(19)	59	(24)
Other CVD	3	(23)	2	(25)	5	(24)	46	(34	41	(37)	87	(36)
Stroke	1	(8)	0	(0)	1	(5)	10	(7)	9	(8)	19	(8)
GORD	0	(0)	2	(25)	2	(10)	23	(17)	15	(14)	38	(16)
Type 2 DM	1	(8)	0	(0)	1	(5)	26	(19)	10	(9)	36	(15)
Osteoarthritis	1	(8)	0	(0)	1	(5)	10	(7)	12	(11)	22	(9)
Osteoporosis	2	(15)	0	(0)	2	(10)	12	(9)	11	(10)	23	(9)
Cancer	5	(38)	1	(13)	6	(29)	32	(24)	20	(18)	52	(21)
Depression	0	(0)	1	(13)	1	(5)	16	(12)	10	(9)	26	(11)
Anxiety	0	(0)	1	(13)	1	(5)	15	(11)	11	(10)	26	(11)
Other mental health	0	(0)	0	(0)	0	(0)	10	(7)	7	(6)	17	(7)

Abbreviations: NZ = NZ; MI = myocardial infarction; IHD = ischaemic heart disease; CVD = cardiovascular disease; GORD = gastro-oesophageal reflux disease; DM = diabetes mellitus.

Medications

Of 340 records, 308 listed the patient’s medications at the index admission. There was a median of 10 medications per person at discharge from the index admission (range 0-25). 115 patients (37%) were taking 5-9 concurrent medications, while 167 patients (54%) took ≥10 concurrent medications. The most common inhaled therapy was LABA, LAMA and ICS (35%), followed by LABA and ICS (21%) then ICS alone (10%), while 23% were recorded as not on any of these three treatments.

While the indication (dyspnoea or pain) was not routinely available from the discharge summary, 24 (8%) patients took short-acting opioids alone, 8 (3%) patients took long-acting opioids alone, and 18 (6%) patients took both. 37 (12%) patients took intranasal midazolam and 11 (4%) patients took clonazepam drops. A total of 37 (12%) patients had domiciliary oxygen at their index admission.

Uptake of interventions

Eighty (24%) patients were current smokers at their index admission and documentation in the notes of smoking cessation advice being given was available for 43 (13%). Rates of smoking and cessation advice were similar across the region.

Of 169 patients in Dunedin, 26 (15%) were recorded as having been offered pulmonary rehabilitation in the 12 months prior to, or 3 months after the index admission. Of these, 5 (3%) completed the programme, 4 (2%) were either currently in the programme or no information was available as to whether they had completed the programme, and 17 (10%) had declined the programme or not completed it. Patients were less likely to be offered pulmonary rehabilitation if they lived in Invercargill or outside of Dunedin and Invercargill or if they lived more than 20 km away from the hospital where it took place (odds ratio for being offered pulmonary rehabilitation compared to Dunedin of 0.10 [95% CI 0.02 to 0.43, p = 0.002] with no offers to those outside of the two cities, and odds ratio of 0.12 for those living further away [95% CI 0.02 to 0.93, p = 0.04] respectively). All patients offered pulmonary rehabilitation had 2 or more co-morbidities. We observed no effects of age, sex, Māori ethnicity, or deprivation decile on whether patients were offered pulmonary rehabilitation.

In most cases, where notes were made of reasons for declining pulmonary rehabilitation, these were related to comorbidities (for example, a heart condition, urinary incontinence or being generally unwell) and one patient was too busy. Eight patients were referred to suitable local exercise classes after they declined pulmonary rehabilitation. We observed no effect of age, sex, Māori ethnicity, number of comorbidities, distance between home and hospital or deprivation decile on whether patients declined pulmonary rehabilitation. Records of pulmonary rehabilitation were unavailable for patients outside Dunedin at the time of this review.

Of 169 patients in Dunedin, 66 (39%) were recorded as having a discussion regarding advance care planning. Advance care planning was not routine for clinical staff in hospitals outside of Dunedin at the time of this review.

Discussion

This retrospective case note review was conducted to describe the health care utilisation of 340 people with severe COPD served by the Southern DHB in NZ. We found that geometric mean (95% CI) length of stay was 3.0 (2.7 to 3.4) days for 241 urban hospital admissions and 4.0 (2.9 to 5.4) days for 47 rural hospital admissions. A greater proportion of patients were from areas of greater deprivation but hospital length of stay was increased in patients from the least deprived areas. Patients had multiple comorbidities and polypharmacy was common: there was a median of 4 (range 0-16) comorbidities and 10 medications (range 0-25) per person. Of 169 cases in Dunedin, where data were available, 26 (15%) were recorded as having been offered, and 5 (3%) completed, pulmonary rehabilitation at or in the year prior to the index admission.

A major strength of this study was that we undertook a comprehensive and detailed review of the health care utilisation of all patients presenting to hospital with COPD in 2015 in a NZ region of low population density. The issues faced by the Southern DHB in provision of care are generalisable to other regions of NZ and other countries with similar health care systems and regions of low population density. A further strength of the study is that all cases were confirmed by spirometry, although this may have led to bias in excluding cases of patients less likely to perform spirometry (potentially those with worse access to health care such as patients from socially deprived or rural areas). The study was limited by incomplete data (such as for recording of pulmonary rehabilitation) and apparent differences in reporting of patient contacts (such as for palliative care) across the Southern Region.

A greater proportion of patients at the index admission were from quintiles of greater deprivation, which is consistent with previous work in NZ [6]. Once admitted, however, these patients had shorter lengths of stay. It is not clear why this was the case but it may suggest hospital attendance with milder exacerbations. This might partly be explained by the lack of expense and relative ease of access at the emergency department compared with the fee for service associated with GP appointments and the frequent wait times in the local NZ health care system, although reasons for choosing to attend an emergency department can be complex [15,16]. Mean length of stay was not found to be different in urban and rural hospitals and this is consistent with findings from previous Australian and Canadian studies [17,18].

People with COPD are known to have a high prevalence of associated comorbidities and health care use rises in association with an individual’s total number of comorbidities [5,19]. This group of patients with severe COPD had multiple comorbidities and polypharmacy was common. They spent a considerable amount of time in hospital because of their co-morbidities: in the year prior to index admission, patients spent a mean of 9.0 days in hospital, and nearly half of these days were unrelated to an acute exacerbation of COPD. Similarly, nearly half of repeat admissions within 30 days were related to a cause other than COPD or respiratory tract infection. An admission for acute exacerbation of COPD is therefore an important opportunity to optimise care for a patient’s comorbidities as well as treating the exacerbation itself. Globally, nationally and locally, it is recognised that addressing comorbidities requires a model of care that better integrates primary and secondary health care decision-making [13,20–22].

Where data were available, the completion rate for pulmonary rehabilitation of 3% was very low but consistent with previously reported NZ data showing that less than 1% of patients with COPD took part in pulmonary rehabilitation classes each year [10]. Pulmonary rehabilitation is effective in improving quality of life, exercise capacity and reducing hospital admissions but there is limited literature on its provision in rural areas of NZ and internationally [7,23]. Routine monitoring and evaluation of the offering, uptake and completion of pulmonary rehabilitation among COPD patients in the Southern Region could improve our understanding of why completion rates are so low. In this study, we showed that patients in the Dunedin area were less likely to be offered pulmonary rehabilitation if they lived rurally and if they lived further from the hospital (where the only available pulmonary rehabilitation classes were held). This is consistent with both recent qualitative work in the Southern Region indicating lack of geographic availability as a barrier to access and also previous literature showing that transport and travel problems have a major impact on patients’ ability to attend and complete classes [14,24]. Recently published Australasian guidelines recommend that home-based pulmonary rehabilitation be offered as an alternative to hospital-based pulmonary rehabilitation [7]. There is some recent evidence that this approach is effective in NZ and it might be more acceptable to patients living in rural areas and others for whom transportation and travel to urban centres presents a problem [25].

Conclusions

Our study showed that there were deficits in the care provided to patients with severe COPD. Provision and uptake of non-pharmacological interventions such as pulmonary rehabilitation was suboptimal and unevenly distributed across the region. It was also clear that this cohort of patients has multiple comorbidities and they require a holistic approach to their care. In such a dispersed population, integrated models of care, with better linkages between primary and secondary care, might improve provision and equity of health care delivery to patients with COPD [26,27]. The findings from this study have informed a programme of work to improve the care for patients with COPD in the Southern Region. Further planning and research is needed to develop and assess changes to routine data collection and monitoring, and to develop effective and cost-effective pulmonary rehabilitation services and integrated models of care.

Declaration of interests

The authors declare that they have no competing interests.

Additional information

Funding

We should like to thank Andrew Gray, Consulting Biostatistician, Division of Health Sciences, University of Otago, for his input into the statistical analysis. The study was funded by the Health Research Council of New Zealand (HRC). Research Partnerships for New Zealand Health Delivery Grant 15/655).