Diagnosis of Pulmonary Malignancy after Hospitalization for Pneumonia

Article Outline

• Abstract
• Methods
  • Inclusion and Exclusion Criteria
  • Data
  • Outcomes
  • Statistical Analyses
• Results
• Discussion
• References
• Copyright

Abstract 

Background

Many physicians recommend that patients receive follow-up chest imaging after the diagnosis of pneumonia to ensure that a pulmonary malignancy is not missed. However, there is little research evidence to support this practice. Our aims were to assess the frequency of the diagnosis of pulmonary malignancy, and to identify risk factors for pulmonary malignancy following hospitalization for pneumonia.

Methods

By excluding patients with a prior diagnosis of pulmonary malignancy, we examined the incidence of a new pulmonary malignancy diagnosis in inpatients aged ≥65 years with a discharge diagnosis of pneumonia in fiscal years 2002-2007, and at least 1 year of Department of Veterans Affairs outpatient care before the index admission.

Results

Of 40,744 patients hospitalized with pneumonia, 3760 (9.2%) patients were diagnosed with pulmonary malignancy after their index pneumonia admission. Median time to diagnosis was 297 days, with only 27% diagnosed within 90 days of admission. Factors significantly associated with a new diagnosis of pulmonary malignancy included history of chronic pulmonary disease, any prior malignancy, white race, being married, and tobacco use. Increasing age, Hispanic ethnicity, need for intensive care unit admission, and a history of congestive heart failure, stroke, dementia, or diabetes with complications were associated with a lower incidence of pulmonary malignancy.

Conclusion

A small, but clinically important, proportion of patients are diagnosed with pulmonary malignancy posthospitalization for pneumonia. Additional research is needed to examine whether previously undiagnosed pulmonary malignancies might be detected at admission, or soon after, for those hospitalized with pneumonia.

Keywords: Cancer, Incidence, Pneumonia

Pneumonia is the second leading cause of hospitalization (after childbirth) in the US, with >1.2 million hospitalizations in 2006.¹ In the US, pneumonia (along with influenza) is the 8th leading cause of death and the number one infectious cause of death.² In the last 2 decades, patients aged 65-84 years have shown a 20% increase in the number of pneumonia-related hospitalizations.

Although many physicians recommend that patients receive follow-up chest imaging after the diagnosis of pneumonia to ensure that a pulmonary malignancy is not missed, there is little research evidence to support this practice. Few studies have examined the incidence of diagnosed pulmonary malignancy after hospitalization for pneumonia.³, ⁴ These studies showed that only a small number (<2%) of pneumonia patients received a new diagnosis of pulmonary malignancy, but they were limited by a short follow-up period (<3 months after initial hospitalization) and inclusion of very young patients (>18 years of age) who are at low risk for pulmonary malignancies.

The aims of our study were to assess the frequency of the diagnosis of pulmonary malignancy, either primary or metastatic, after hospitalization for outpatient-acquired pneumonia in subjects ≥65 years of age, and to identify risk factors for posthospitalization pulmonary malignancy.

Back to Article Outline

Methods 

For this study we used data from the administrative databases of the Department of Veterans Affairs (VA) Health Care System. These databases are the repositories of clinical data from all of the VA hospitals and outpatient clinics.⁵ The Institutional Review Board of the University of Texas Health Science Center at San Antonio approved this study under expedited review.

Inclusion and Exclusion Criteria 

Patients included in this study:

As our primary aim was to examine the incidence of new pulmonary malignancy, we excluded patients with preexisting pulmonary malignancies (assessed October 2000 forward) from our analyses. We assessed only the first hospitalization for pneumonia during the study period and excluded patients with an admission for pneumonia in the previous 12 months (1-year cleanout period).

Data 

We used demographic, utilization, and comorbidity data from the National Patient Care Database, pharmacy data from the VA Decision Support System National Data Extracts and Pharmacy Benefits Management, and vital status information from VA's Vital Status file, which incorporates data from veterans' death benefits claims, inpatient deaths, Medicare Vital Status files, and the Social Security Administration death master file. Encrypted patient identifiers linked the information across these databases.

We obtained demographic information (age, sex, race, marital status) from inpatient and outpatient data. Race categories included white, black, Hispanic, and other/unknown. International Classification of Diseases, 9^th Revision (ICD-9) codes for tobacco use (305.1, V15.82), smoking cessation clinic use, or use of medications for the treatment of nicotine dependence (eg, Zyban, nicotine replacement, or varenicline) were used to infer tobacco use.

We also obtained information on comorbid conditions from inpatient and outpatient administrative data. Alcohol abuse was defined by ICD-9 codes 291, 303, 305.0, and illicit drug use by ICD-9 codes 292, 304, and 305, excluding 305.1. We used Charlson's comorbidity methodology to classify other preexisting comorbid conditions, both individually and as a composite score.⁷, ⁸ Charlson's comorbidity system includes 19 comorbid conditions, which are classified using ICD-9 codes from prior outpatient and inpatient codes.⁹

Outcomes 

We used a new diagnosis of either primary lung cancer or pulmonary metastasis after index hospitalization as the primary outcome for this study. ICD-9 codes were used to identify primary or metastatic pulmonary cancers. These ICD-9 codes included 162, 163, 165, 165.0, 165.8, 165.9, 176.4, 197.0, 197.1, 197.2, 197.3, 198, 199, 199.0, 199.1, 230, 231, 231.1, 231.2, 231.8, 231.9, 234, 235, 235.6, 235.7, 235.8, 235.9, and 238. Prior research has validated the use of ICD-9 codes to identify pulmonary malignancies.¹⁰

Secondary outcomes included 30- and 90-day mortality, long-term survival, and length of hospital stay for the index admission. Mortality was assessed through October 1, 2007 using the VA vital status file. Previous studies have demonstrated that this methodology has a sensitivity of ∼98% for veterans' deaths.¹¹

Statistical Analyses 

Bivariate statistics were used to test the association of sociodemographic and clinical characteristics with all-cause 30-day mortality. Categorical variables were analyzed using the chi-squared test, and continuous variables were analyzed using Student's t test. Due to the number of statistical tests and the large sample size of this database, we set statistical significance at P ≤.001.

To evaluate risk factors for a postpneumonia hospitalization diagnosis of pulmonary malignancy, baseline patient sociodemographic and clinical characteristics were used as independent variables in an extended Cox model.¹² The baseline variables included comorbid conditions that were significantly associated with pulmonary malignancy, age group at admission (65-74, 75-84, 85+ years), race/ethnicity, intensive care unit (ICU) admission, tobacco use, and marital status. Due to failure of the proportional hazards assumption in the standard Cox models, 2 variables, age group at admission and history of metastatic cancer, were included as time-varying covariates. Therneau et al's method was used to evaluate the proportional hazards assumption,¹³ and our final model satisfied these criteria.

To analyze time-to-diagnosis of pulmonary malignancy, a graph was created using Kaplan-Meier estimated probabilities. To examine the impact of a new diagnosis of pulmonary malignancy on mortality over the entire period of follow-up, we used Cox proportional hazard models to estimate and graph the baseline survivor functions after adjusting for variables associated with pulmonary malignancy, which were stratified into 2 time periods (0-90 days and 90-1500 days). We broke these 2 time periods into separate graphs due to an inversion of risk at ∼90-100 days.

All analyses were performed using STATA 9 (StataCorp LP, College Station, Tex).

Back to Article Outline

Results 

We identified 40,744 patients who met the inclusion criteria and who did not have a diagnosis of pulmonary malignancy before the pneumonia hospitalization. The mean age was 77.7 years, SD 6.8 years; 53.0% of subjects were married and 98.1% were male. In this cohort, 10.5% were black, 66.6% were white, 1.5% were Hispanic, 4.6% were other, and 16.8% were of unknown race/ethnicity. Regarding mortality, 12.9% (n=5270) of the subjects died within 30 days of admission, and 20.7% (n=8451) died within 90 days of admission. In our cohort, 3760 (9.2%) patients were diagnosed with pulmonary malignancy after hospitalization for pneumonia.

Table 1 shows descriptive demographics and comorbid conditions by subsequent diagnosis of pulmonary malignancy. Factors significantly associated with increased risk of diagnosis of pulmonary malignancy in univariate analyses included lower age, white race, being married, current tobacco use, chronic obstructive pulmonary disease, prior malignancy and metastatic solid tumor (both excluding prior pulmonary malignancies). Factors associated with lower risk of diagnosis of pulmonary malignancy include being black or Hispanic, requiring ICU admission during the hospitalization, requiring mechanical ventilation, congestive heart failure, prior stroke, dementia, diabetes with complications, diabetes without complications, and renal disease.

Table 1. Characteristics of Patients (n=40,744) Hospitalized with Pneumonia by Diagnosis of Pulmonary Malignancy

⁎Excluding prior pulmonary malignancies.

Median time to diagnosis of pulmonary malignancy was 297 days (mean 452.1, SD 442.7), and only 27% were diagnosed within 90 days of admission. Figure 1 depicts time to diagnosis of pulmonary malignancy.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  Time to new diagnosis of pulmonary malignancy for patients aged ≥65 years hospitalized with pneumonia.

Table 2 presents the results of the extended Cox model. In this model, factors significantly associated with postpneumonia pulmonary malignancy included history of chronic obstructive pulmonary disease (hazard ratio [HR] 1.12, 95% confidence interval [CI], 1.04-1.20), any prior malignancy (HR 1.69, 95% CI, 1.56-1.83) or metastatic solid tumor (HR 3.33, 95% CI, 2.10-5.12), (both excluding those with prior pulmonary malignancy), white race (HR 1.15, 95% CI, 1.06-1.24), being married (HR 1.10, 95% CI, 1.03-1.17), and current tobacco use (HR 1.37, 95% CI, 1.28-1.46). Increasing age group (HR 0.88, 95% CI, 0.82-0.95), Hispanic ethnicity (HR 0.60, 95% CI, 0.44-0.82), need for ICU admission (HR 0.48, 95% CI, 0.43-0.55), having a history of congestive heart failure (HR 0.92, 95% CI, 0.85-0.99), stroke (HR 0.80, 95% CI, 0.74-0.88), dementia (HR 0.53, 95% CI, 0.43-0.65), renal disease (HR 0.88, 95% CI, 0.79-0.88), and diabetes with complications (HR 0.88, 95% CI, 0.87-0.99) were associated with a lower incidence of diagnosis.

Table 2. Results of the Extended Cox Model for Time Until Diagnosis of Pulmonary Malignancy

⁎Excluding prior pulmonary malignancies.

As a sub-analysis, we excluded all subjects who died within 90 days of hospital admission and re-ran the survival analyses. We found that factors still associated with new diagnosis of pulmonary malignancy included current tobacco use (HR 1.18, 95% CI, 1.15-1.23), any malignancy (HR 1.13, 95% CI, 1.08-1.18), metastatic solid tumor (HR 7.46, 95% CI, 4.9-11.4), chronic obstructive pulmonary disease (HR 1.12, 95% CI, 1.08-1.16), congestive heart failure (HR 1.16, 95% CI, 1.12-1.20), and male sex (HR 1.26, 95% CI, 1.11-1.41).

Figure 2, Figure 3 are plots of survival time by whether patients received posthospitalization diagnosis of pulmonary malignancy. Figure 2, which covers 0-90 days postadmission, demonstrates that patients with a subsequent diagnosis of pulmonary malignancy had significantly lower short-term mortality as compared with those who did not (P <.001). Figure 3, which covers 90+ days after the pneumonia admission, shows that patients with a diagnosis of pulmonary malignancy had significantly higher long-term mortality as compared with those who did not (P <.0001).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2. 

  Proportion surviving within the first 90 days by posthospitalization diagnosis of pulmonary malignancy in patients with pneumonia (P <.0001).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 3. 

  Proportion surviving after 90 days by posthospitalization diagnosis of pulmonary malignancy in patients with pneumonia (P <.0001).

Back to Article Outline

Discussion 

We found that a small, but clinically important, number of elderly patients with pneumonia are diagnosed with posthospitalization pulmonary malignancy. Our results support the concept of follow-up imaging for elderly patients with pneumonia. However, as this is a retrospective study, we are unable to comment on the appropriate time frame or imaging modality (eg, chest radiograph, computed tomography of the chest).

In our study we found that there were significant differences in several comorbid conditions and demographics among those subjects who were subsequently diagnosed with pulmonary malignancies and those who were not. Many of these are quite clearly related to increased incidence of pulmonary malignancy (eg, tobacco use, chronic obstructive pulmonary disease, history of cancer or metastatic cancer). However, others are not as clear (eg, history of stroke, renal disease, or dementia), and we hypothesize that these are ”protective” against pulmonary malignancy due to increased mortality associated with these conditions so that many of these patients die before developing clinically significant pulmonary malignancies. In addition, we found that increasing age in our cohort was associated with lower incidence of pulmonary malignancy. This corresponds to data from the Surveillance Epidemiology and End Results cancer registries that reported the median age of diagnosis is 70 years of age, which is on the lower side of the age range of our cohort.

Interestingly, we found significant differences in outcomes (30-day and 90-day mortality, long-term survival, and length of stay) between the patients with postpneumonia pulmonary malignancy and those without, and in factors such as ICU admission. It is noteworthy that patients subsequently diagnosed with lung cancer had improved survival for the first 90 days postpneumonia hospitalization, however, lower survival after 90 days. It appears that patients with pulmonary malignancy have survivals <2 years, which is consistent with previous literature.¹⁴ This suggests that there are significant differences between the 2 groups even at the time of pneumonia hospitalization, and that even though there is a substantial delay until the initial diagnosis of pulmonary malignancy, the pulmonary malignancy is at least partially responsible for their initial hospitalization.

In this study we found a much higher incidence of pulmonary malignancy as compared with previous studies. Marrie³ found, in a prospective study of 1269 patients with community-acquired pneumonia, that only 25 (1.97%) had lung cancer, while Holmberg and Kragsbjerg⁴ found that 1.3% of patients with community-acquired pneumonia had a previously undiagnosed pulmonary malignancy. Both studies had significant limitations, including a short and unclear follow-up period (<3 months), and inclusion of very young patients (>18 years of age), who are at very low risk for pulmonary malignancies. We believe that the differences in the results between these studies and ours were due to our older population, which is at much higher risk for pulmonary malignancy and the much longer follow-up (up to 5 years).

Our study has several limitations. Although our study was a large database analysis and subject to the recognized limitations of such studies, we carefully assembled our cohort from complete patient discharge data to avoid ascertainment bias. However, due to the large number of subjects in our study routine, definitions of statistical significance (P <.05) are meaningless. Therefore, we based our decisions on which variables to include in the survival analysis on clinical relevance. Our sample was predominantly men, due to our use of VA administrative data, and further research is needed to examine these issues in women. Although we relied upon ICD-9 codes to define both pneumonia and pulmonary malignancies, previous studies have validated both.¹⁰, ¹⁵, ¹⁶, ¹⁷ In addition, due to the design of the primary study, we examined only patients ≥65 years of age. Future studies also should examine other age groups. Also, we were unable to examine severity of illness at presentation using previously validated measures such as the Pneumonia Severity Index,¹⁸ due to the lack of several variables (eg, systolic blood pressure). Next, data on when and where radiographic testing was done are not available in these databases, so we are unable to examine these issues. Finally, as this is not a prospective cohort study, we are unable to identify when these malignancies would be able to be first identified. Therefore, we are unable to address whether there is a true “physiologic” delay between hospitalization and diagnosis or, as we suspect, many of these patients did not get repeat chest imaging until they developed additional symptoms. Further research is needed to determine what would be the appropriate timing of follow-up testing, and what that testing would be.

In conclusion, our study finds that a small, but clinically important, number of elderly patients are diagnosed with pulmonary malignancy posthospitalization for pneumonia. Posthospitalization follow-up is important for those previously hospitalized with pneumonia, especially for those at higher risk for malignancy. Further research is needed to examine other groups, such as those <65 years of age and women, and to examine whether strategies to detect previously undetected pulmonary malignancy at time of admission, or soon after, are associated with improved outcomes for patients hospitalized with pneumonia.

Back to Article Outline

References 

1. Pneumonia most common reason for hospitalization. AHRQ News and Numbers. Rockville, MD: Agency for Healthcare Research and Quality; 2008;
   • View In Article
2. Kung HC, Hoyert DL, Xu JQ, Murphy SL. Deaths: final data for 2005. Natl Vital Stat Rep. 2008;56:1–120
   • View In Article
3. Marrie TJ. Pneumonia and carcinoma of the lung. J Infect. 1994;29:45–52
   • View In Article
   • MEDLINE
   • CrossRef
4. Holmberg H, Kragsbjerg P. Association of pneumonia and lung cancer: the value of convalescent chest radiography and follow-up. Scand J Infect Dis. 1993;25:93–100
   • View In Article
   • MEDLINE
   • CrossRef
5. Brown SH, Lincoln MJ, Groen PJ, Kolodner RM. VistA—U.S. Department of Veterans Affairs national-scale HIS. Int J Med Inform. 2003;69(2-3):135–156
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (427 KB)
   • CrossRef
6. Meehan TP, Fine MJ, Krumholz HM, et al. Quality of care, process, and outcomes in elderly patients with pneumonia. JAMA. 1997;278:2080–2084
   • View In Article
   • MEDLINE
7. Charlson ME, Pompei P, Ales KL, MackKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–383
   • View In Article
   • MEDLINE
   • CrossRef
8. Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol. 1994;47:1245–1251
   • View In Article
   • MEDLINE
   • CrossRef
9. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613–619
   • View In Article
   • MEDLINE
   • CrossRef
10. Thomas SK, Brooks SE, Mullins CD, et al. Use of ICD-9 coding as a proxy for stage of disease in lung cancer. Pharmacoepidemiol Drug Saf. 2002;11:709–713
    • View In Article
    • MEDLINE
    • CrossRef
11. Sohn MW, Arnold N, Maynard C, Hynes DM. Accuracy and completeness of mortality data in the Department of Veterans Affairs. Popul Health Metr. 2006;4:2
    • View In Article
12. Therneau TM, Grambsch PM. Modeling Survival Data: Extending the Cox Model (Statistics for Biology and Health). New York: Springer; 2000;
    • View In Article
13. Therneau TM, Grambsch PM, Fleming TR. Martingale hazards regressions and the analysis of censored survival data. Biometrika. 1990;77:147–170
    • View In Article
    • CrossRef
14. van Rens MT, de la Riviere AB, Elbers HR, van Den Bosch JM. Prognostic assessment of 2,361 patients who underwent pulmonary resection for non-small cell lung cancer, stage I, II, and IIIA. Chest. 2000;117:374–379
    • View In Article
    • MEDLINE
    • CrossRef
15. Marrie TJ, Durant H, Sealy E. Pneumonia—the quality of medical records data. Med Care. 1987;25:20–24
    • View In Article
    • MEDLINE
    • CrossRef
16. Whittle J, Fine MJ, Joyce DZ, et al. Community-acquired pneumonia: can it be defined with claims data?. Am J Med Qual. 1997;12:187–193
    • View In Article
    • MEDLINE
    • CrossRef
17. Aronsky D, Haug PJ, Lagor C, Dean NC. Accuracy of administrative data for identifying patients with pneumonia. Am J Med Qual. 2005;20:319–328
    • View In Article
    • MEDLINE
    • CrossRef
18. Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med. 1997;336:243–250
    • View In Article
    • MEDLINE
    • CrossRef